Joy & Curiosity #59

Interesting & joyful things from the previous week

There’s one thing from the second Acquired episode on Google that I keep thinking of. They mention how, in 2019, Chrome started to hide the protocol in the address bar. No more http and https. Both Acquired hosts agree that this was obviously the right choice. No user should’ve ever had to worry about typing http colon slash slash.

And, man, now, six years later, I agree. I think it was the right call. And I think they nailed it. Lots of small details: you don’t see the protocol by default, but when you hit Cmd-L to select the URL the protocol is shown and selected too. If you click on it though, you don’t see the protocol, but if you click and copy, it is copied too. And so on.

But what I keep thinking of: it was the right call and I, someone who likes to think of himself as a person who has some product taste, 100% would not have made it. It haunts me.

• Very, very impressive and very, very good: Build Your Own Database, a “step-by-step guide to building a key-value database from scratch”, which really undersells it. To use some words that I heard as a kid when grown-ups would talk about the Information Superhighway: it’s interactive, it’s visual, it’s multi-media. Okay, maybe it’s not multi-media, who knows, but it’s very good.

• Vicki Boykis on mastery: “There is no single element that comprises this quality of care, the striving towards excellence and mastery. But, as Bernoulli said when he received an anonymous solution to the brachistochrone problem that turned out to be Isaac Newton, ‘I recognize the lion by its claw.’ […] I recognize the claw of the lion in software like Redis, cURL, uv, Ghostty, sqlite, llama.cpp - software that is elegant, well-built, considered and thoughtful. Software that is joyful to use. Software that helps me. I want to write software like this, and I want to use software like this, and I want us as programming people to be incentivized to value the process that creates software like this.” Loved it. Even though I disagree with the pessimistic assessment that “[we] are being overrun by mediocrity and sloppiness” — reminds me of what “real programmers” said about “web developers” 15 years ago.

• Linked in Vicki’s post is this wonderful article by Jacob Kaplan-Moss: “You were hired to write code. Many developers make the mistake and think that their job stops there. That’s not true. In fact, you have two jobs: (1) Write good code. (2) Be easy to work with.” (That’s exactly what I meant to say with the last sentence here.)

• Some described it as “glazing”, others made fun of some phrases in it, but, I don’t know man, I really enjoyed reading this: The New World. It’s a long-form profile of Joshua Kushner, who I didn’t even know existed, and who’s the founder of Thrive Capital, which I also didn’t know. But it’s also about the Holocaust, and the American Dream, and a family, and it mentions Linus Lee, and mini-profiles other people at Thrive, and goes into the The Week Sam Altman Was Fired, and how Thrive invested in GitHub, and how they invested in Stripe, and… yes, I really enjoyed reading it.

• Wonderful: Programming With Less Than Nothing.

• As are the Aphyr posts on interviews. Often think of them, believe it or not.

• My teammate Nicolay: Don’t delegate thinking, delegate work. “The bottleneck now isn’t typing, it is understanding and problem solving, it is the thinking that happens before and after the code appears. What problem am I actually solving? Does this approach make sense given what already exists? Will this make the codebase more comprehensible or less? Those questions don’t get faster to answer just because the code appears faster. If anything, they get harder, because now you’re reviewing 500 lines instead of 50.”

• Did you wake up this morning wishing you could read something about salamis, mold, humidifiers, fridges, Home Assistant, and “food industry framework for identifying what can go wrong and how to monitor it”? I got something for you: Designing Software for Things that Rot. This was great.

• So, last week we launched Amp Free. Someone (but not one of us) posted us to Product Hunt. And ever since then my inboxes — email, LinkedIn, Twitter — have been full of people offering me Product Hunt-upvote-armies and other scammy-sounding services related to Product Hunt. In a tweet, I wondered whether Product Hunt has become a zombie town. In the replies, people told me: yes, yes it is. Some even said Product Hunt might be the biggest argument in favor of the Dead Internet Theory and then someone linked me to this post on the “brutal reality” of a Product Hunt launch and, well, I can’t dispute it.

• Turn off Cursor, turn on your mind. I agree that the risk is there. I’ve been there. But I don’t agree with the blanket statement. Not all code is worth the understanding.

• Scripts I wrote that I use all the time by Evan Hahn. I feel inspired and lazy at the same time reading this. I have maybe three or four homemade scripts that I use semi-regularly and I somehow really wish that was different. Time to change it. (Make sure to click that second link too, that photo is worth it.)

• Patrick sent this to me, asking whether I’ve seen it. I hadn’t. I constantly forget that George Saunders (George Saunders!) has a Substack. And, wow, am I glad he reminded me of it. This post is wonderful: A Tough Question Indeed. Here: “Only later (years later) did I find out that Celine was a bit of a turd – a collaborationist and an anti-Semite. But…those five pages had changed the trajectory of my life, regardless of their source. Would I ‘unread’ them, if I could? Absolutely not. So, it seems to me, two thoughts can co-exist: 1) I like this writing, and 2) I don’t like the person who wrote it.” What a wonderful writer.

• In case you’ve never read anything by George Saunders, maybe start here: what writers really do when they write. It’s… wonderful (and, yes, I know, I overuse that word, but I had “divine” here and then I thought: what do I know about the divine? So, wonderful it is, because it is.)

• A Simon Willison talk turned into blog post: Living dangerously with Claude. Something that I never considered until reading it here and that blew my mind: “In YOLO mode you can leave Claude alone to solve all manner of hairy problems while you go and do something else entirely. I have a suspicion that many people who don’t appreciate the value of coding agents have never experienced YOLO mode in all of its glory.” Can that be true? People have tried coding agents but only in the mode where they accept or reject every change and command? Wow.

• OpenAI released Atlas, their own browser. The onboarding has some delightful moments. Maybe I’m totally off here, but to me this looks like the work of all the ex-Meta people that have joined OpenAI. I’ve played around with it a bit, but not too much. Very curious to see where this all goes and whether this means “computer use” (in the sense of moving a cursor across a screen) is dead and that agents/models will live on the web.

• Man, Greenland sounds bad: “I have been to the jungles of Vietnam, the swamps of Florida and the Canadian countryside. This was beyond anything I’ve ever experienced. There are bugs in my mouth, ears, eyes and nose almost immediately.”

• Then again, OpenAI also acquired Sky and the PR statement contains this: “Sky’s deep integration with the Mac accelerates our vision of bringing AI directly into the tools people use every day.” Why integrate with the Mac if you care about the web? Wild times.

• This was interesting: Everything Is Television, by Derek Thompson. “Television speaks to us in a particular dialect, Postman argued. When everything turns into television, every form of communication starts to adopt television’s values: immediacy, emotion, spectacle, brevity. In the glow of a local news program, or an outraged news feed, the viewer bathes in a vat of their own cortisol. When everything is urgent, nothing is truly important. Politics becomes theater. Science becomes storytelling. News becomes performance. The result, Postman warned, is a society that forgets how to think in paragraphs, and learns instead to think in scenes.”

• David Friedman built Doomscrolling, the game. “As readers know, I’m not a coder, but I enjoy how vibe coding lets me turn an idea into something real. So naturally, I turned to vibe coding for this. It didn’t work. [...] But then GPT-5 came out a few weeks ago, and I tried again to see how much better it might be at coding. In just two hours I had a very good prototype.” So far, so familiar, but this here is very interesting and real: “If you’ve ever tried to work with AI, you’ve likely run into a roadblock where you’re describing something over and over and it’s simply not getting it. ‘No, don’t do it that way, for the umpteenth time, do it this way.’ So I simplified things. I had the AI set up simple ‘labs,’ standalone test pages where we could work on different designs, using the style from the game.” And that’s exactly the kind of code I had in mind when I wrote this.

• AI is Making Us Work More: “Where feeling tired used to be a signal to rest, now it’s a sign of weakness. Every break you take feels like a gap in your potential productivity.” Yes. Something changed this year in software and I’d argue that if you haven’t felt it, it’s not because it didn’t happen, but because it hasn’t reached you yet.

• Geoffrey Litt: “Personally, I’m trying to code like a surgeon. A surgeon isn’t a manager, they do the actual work! But their skills and time are highly leveraged with a support team that handles prep, secondary tasks, admin. The surgeon focuses on the important stuff they are uniquely good at. My current goal with AI coding tools is to spend ~100% of my time doing stuff that matters.”

• “We’re past the ‘wow, it writes code’ phase. The interesting work now is shaping how these assistants fit real workflows: deciding when to be fast and when to be careful, what context to include, and how to keep humans in the loop without overwhelming them.”

• Simon Willison again: “I actually don’t think documentation is too important: LLMs can read the code a lot faster than you to figure out how to use it.” And other thoughts on how to make agents productive. My hypothesis: the best codebase for agents doesn’t look like the best codebase for humans. At least not short-term.

• John Collison in The Irish Times: “This phenomenon is not just Irish. Around the developed world, power has shifted from politicians to officials. The book Why Nothing Works by Marc Dunkelman divided US political history into the period before 1970 and the period after. The period before 1970 it said was focused on building capacity, the period afterwards on constrainingcapacity: ‘If progressivism had once been focused on building up centralized institutions, the new goal was to tear them down.’”

• “quit brainrot.”