Joy & Curiosity #83
Interesting & joyful things from the previous week
This is a time of great technological change. You could even wring a “once in a lifetime” out of me . Many times per week now I say to either myself or someone who just shared some news: this is crazy, man.
The numbers, the pace, the demand, the bottlenecks shifting, the new capabilities emerging, and, man, the predictions. The predictions. AI will do that, AI will do this, in the future we’ll do all of this and none of that, but surely this will still be that and that thing will be the most important thing.
I’ve done it too, of course. I’ve predicted quite a few things in past issues of this newsletter and, hey, yes, I was right a few times. And so were others.
But we’re talking about technological progress here and that is very hard to predict, especially its second-order effects. So, as you read through the things I shared below, I want you to keep the following quote in mind, because it’s been stuck in mine for many weeks now and I found it helpful to carry around with me:
He did not create a world that went as he wanted, but he created a world that went well. We have many examples of that. Trains and bicycles come in, and we get feminism because it’s easier for people, especially women, to move freely and independently. They can organize. They can mobilize. We get suffragettes. Did the inventor of the train intend for there to be women’s liberation? No. Did it go the way he imagined? No. Did it go well? Yes.
Or consider this:
After the Great War, the Haber-Bosch process was used throughout the world to fix nitrogen on a grand scale. […] It was synthetic fertilizer that enabled Europe, the Americas, China and India to escape mass starvation and consign famine largely to the history books: the annual death rate from famine in the 1960s was 100 times greater than in the 2010s. […] If Haber and Bosch had not achieved their near-impossible innovation, the world would have ploughed every possible acre, felled every forest and drained every wetland, yet would be teetering on the brink of starvation, just as William Crookes had forecast.
That was after the war. Here’s what Bosch and Haber did with their process during the war:
Then in September 1914 Bosch made the famous ‘saltpetre promise’ that he could convert the Oppau plant so that it turned ammonia into nitrate, using a newly discovered iron-bismuth catalyst. He built an even bigger plant at Leuna, producing huge quantities of nitrate and thus probably prolonging the war. Haber, in the meantime, had invented gas warfare, personally presiding over the first chlorine attack at Ypres in March 1915.
Now, who would’ve predicted going from that to that?
Amp’s smart mode now uses Opus 4.7. I think it’s a great model. I now often switch between smart and deep mode. One plans, the other reviews, and vice versa.
Last week I re-read Mike Acton’s Expectations of Professional Software Engineers and, man, is it good. So, so good. If you haven’t, you need to read this right now. This is software engineering in a team, in a company, in a business. Hacking isn’t programming isn’t engineering, but what he describes here, that’s the real thing. And — of course you have to say this, Thorsten — yes: this all still applies when using AI. Maybe even more so. Just like The Basics.
For many, many years I’ve come across strong recommendations to watch this talk by Richard Hamming: You and Your Research. Not considering myself a scientist, I shrugged off those recommendations and never saw it. I can tell you now: that was a huge mistake. This morning, right after waking up, still in bed, I read this transcript, start to end, and let me tell you this: watch the talk or read the transcript! If you’re here, reading this newsletter, I’m certain you will get something out of it. It’s fantastic.
Highly, highly recommend you watch this interview with Dylan Patel on the current state of tokenomics. Really: if you only have a vague idea of what “compute constrained” means, you have to watch this. (Also, the last ten minutes, in which Dylan talks about the optics of the model companies, are kinda separate from tokenomics, but worth it alone.)
Talking of which: “Cursor has also given SpaceX the right to acquire Cursor later this year for $60 billion or pay $10 billion for our work together.” $60 billion (!) now sounds like $60 million did in 2012.
Kevin Kwok’s thoughts on Cursor’s and SpaceX’s partnership are interesting, but I disagree with him on the premise that model and harness have to go hand in hand. I don’t think the causality of the loop is there: Claude 3.5’s ability and eagerness for tool calls was the Urknall of agents. That’s what lead to us to build Amp and Anthropic to build Claude Code.
Bonkers numbers: Google wants to invest up to $60B in Anthropic. The Hacker News comments are interesting.
Justin Jackson is asking: what has technology done to us? I very much don’t agree with the quoted statement of “technology will always do its worst thing” (and neither does Justin, it sounds like.)
It’s cool to care: “Whenever somebody asks why, I don’t have a good answer. Because it’s fun? Because it’s moving? Because I enjoy it? I feel the need to justify it, as if there’s some logical reason that will make all of this okay. But maybe I don’t have to. Maybe joy doesn’t need justification. […] So much of our culture tells us that it’s not cool to care. It’s better to be detached, dismissive, disinterested. Enthusiasm is cringe. Sincerity is weakness. I’ve certainly felt that pressure – the urge to play it cool, to pretend I’m above it all. To act as if I only enjoy something a ‘normal’ amount. Well, fuck that.”
Take some time to play around with ChatGPT Images 2.0. It’s mind-blowing. If they can accurately reconstruct screenshots like, regardless of whether that’s the “image” model part or the “thinking” model part, I think something just shifted. Also, what a sick landingpage.
This was great: What will be scarce? The question that leads to the one in the title is this: “If advanced AI brings material abundance—if machines can produce many if not all forms of human production at very low marginal cost—does economics become irrelevant?” The whole piece is explains the possible mechanisms at play and also answer the question of whether economics will become irrelevant, but even more interesting is the prediction on the future of work: “The economics of structural change tells us that when technology makes one type of production cheap, the economy doesn’t collapse. It transforms. It shifts toward the things that technology can’t make cheap. For AI, those things are exactly the ones where human involvement carries inherent, irreplaceable value.” And that means the “durable jobs will be in the relational sector, where the human element is the product itself.” Or, in other words: “You don’t need to be Picasso. You need to be the person whose involvement makes the product feel like it was made for someone, by someone.”
“A parasite that has been eating people for 3,500 years is about to be wiped off the planet. It infected 3.5 million people in 1986. Last year, it infected 10. And I have not seen it make a single front page.” Believe it or not, but in seventh grade I gave a presentation in biology class on the Guinea worm. Use Google Image search if you’re as brave as I was in seventh grade. Yeah, thought so.
This is from December last year, so the numbers are even crazier now, which makes this even more interesting: Liar’s Valuation. I knew about “take last month’s revenue and multiply by twelve,” but the tiered investment rounds were new to me, and so was the “give heavy discount in year one, but then report year three bookings as ARR.”
The annotated Unicode map. More of this!
Yes, it’s Sky Sports News of all places: “Pressure is a privilege. And if you’re feeling any pressure or the weight of any expectation, you are breathing rare air, that very few of us get to live inside.” Good frame.
Or, as Josh Kushner said: “Every experience is training you for the next one… In order to become king, God didn’t give David a crown, he gave him Goliath.”
Tim Cook is stepping down as Apple’s CEO. This Stratechery reflection was very interesting: Tim Cook’s Impeccable Timing. For example, I had no clue that Apple in China (as in: moving its manufacturing to China) was the work of Cook. For me, Cook will always be the CEO who was at the helm when the M1 shipped, one of most remarkable engineering achievements I’ve witnessed.
Apple’s incoming CEO John Ternus in 2024 in a commencement speech: “At some point in my first year, I found myself at a supplier facility. I was far away from home, it was well past midnight. I was using a magnifying glass to count the number of grooves on the head of this screw, which, remember, lives on the back of the display. And I was arguing with the supplier because these parts had 35 grooves, they were supposed to have 25. I distinctly remember stepping back for a minute and thinking to myself, ‘What the hell am I doing? Is this normal?’ And I thought about it, and I realized it might not be normal, but it’s right. It’s right because I’d already spent months working on that product, and if you’re going to spend that much time on something, you should put in your very best effort. Maybe a customer notices, maybe they don’t, but either way, whenever I saw one of those displays on someone’s desk, it mattered to me to know that my teammates and I had considered everything about it and done the very best job we could.” There’s a lot more good stuff in there. I’m excited.
After probably ten years of using Alfred I switched over to Raycast two years ago and one thing that I’ve sporadically but consistently missed was Alfred’s “Large Type” feature: you type a bit of text hin, hit a shortcut, and boom, the text is now as big as your display. Very helpful when you want to show someone in the room the wifi password, for example. So, this week I thought: surely there’s a Raycast plugin for that? And there is but the text isn’t that large. But guess what, there’s also this: large-type.com. How good is that?
Adam Mastroianni again with some very good writing on capital-S science: Nothing ever dies. It merely becomes embarrassing. I didn’t know that ego depletion doesn’t reproduce! While reading I had to think of Brandolini’s Law: “The amount of energy needed to refute bullshit is an order of magnitude bigger than that needed to produce it.” (In 2015 both Brandolini and I both gave a talk at a Ruby conference in Wrocław, Poland, and we chatted for half an hour at the airport and, not sure exactly why, but I’m oddly proud of that.)
Orson Scott Card, author of Ender’s Game: “Those changes made, I sent it to Ben again. I did not remind him of what he had advised me to do. I merely told him I liked my title, and said, ‘I have addressed your other concerns,’ which was true. I figured he wouldn’t remember what his exact words had been. My answer was a check. [...] Did Ben’s feedback help? Yes — but his specific advice was not right, and I knew it. [...] Editors don’t know more than you about your story. They especially don’t know why they decide to accept or reject stories. YOU have to know what your story needs to be, and take only advice that you believe in.”
Reminded me a lot of Bill Hader on feedback: “When people give you notes on something, when they tell you it’s wrong, they’re usually right. When they tell you how to fix it, they’re wrong.”
exe.dev raised a Series A: “We are building a cloud that makes sense for the current and future state of software development. One that includes the features needed for fast, secure development out of the box. A cloud developers actually enjoy using. We want to revitalize the spirit of projects like early Heroku (though our technology is very different) and ship features that bring you joy.” (Not to take away from this announcement, hence the parenthetical: the impact Heroku had on a certain generation of programmers working on developer tooling is hard to overstate. I bring it up a lot, and so do my teammates who are close my age and worked with web technologies in the early 2010s.) I’m very excited to see what they’ll do! I like using exe.dev a lot.
I also really like David’s personal statement that goes along with the funding announcement: I am building a cloud.
Just a reminder: chat jimmy exists. Try it. You have to. Try it and then imagine what we could do if one of today’s frontier models would run at even half that speed. Send me a letter if you know whether that’s physically impossible.
New Larry David biography is coming out this year. Pretty, pretty, pretty good.
Elad Gil’s Random thoughts while gazing at the misty AI Frontier. Lots of interesting things in there. AI researchers’ distributed IPO, compute constraints, hidden layoffs, and also this bit: “It is not just the model you use, but the environment, prompting, etc you build around it that helps impact your choice. Brand also matters more then many people think. At some point, either one coding model breaks very far ahead, or they stay neck in neck.”
Maggie Appleton: One Developer, Two Dozen Agents, Zero Alignment. I think I see the same future that Maggie sees. And we’re building it at Amp.
That’s a title worthy of a book, not a post, but the content is still fascinating: Fabric is harder than steel. As someone who’s been chasing the perfect t-shirt for years and who has a very deep fascination with “tech shirts” (not company logos, but high-quality shirts made of “functional” textiles), this was very cool. I often wondered: how can car seats be this good for so long? Well, turns out it’s engineering.
Jeff Geerling: New 10 GbE USB adapters are cooler, smaller, cheaper. I could read blog posts like this one five times every day.
I Found It: The Best Free Restaurant Bread in America. This was fantastic. Go read it if you have an hour and want to smile and enjoy some great writing. There are many quote-worthy sentences in there, but I’ll let you read them yourself. Instead, here’s a free bread anecdote. Once upon a time, I was working on a farm in Australia, along with around ten other backpackers. Handful of Germans, handful of French people, two Brits. One day we were sitting around the big table in this “shed” (actually a big house, with a shed-like quality, if you will) we were living in, chit-chatting about stuff. What do you miss the most from home? came up as a question and after someone said that they miss a proper shower and feeling clean for once many of us nodded. Yes, that’d be something. Then someone said: I really, really miss the bread. And everybody, because we’ve all seen and tasted what the Australians call bread, let out a big sigh and said, oh yes, the bread, I miss the bread. And precisely one second later, the room split into two factions and the Germans stared at the French and the French stared at the Germans and both factions, at the same time, said something to the effect of: wait, what the fuck, why do you miss bread, your bread fucking sucks, our bread is good bread, your bread is garbage, shut up. But, sadly, the French wouldn’t see how wrong they were, thinking their long, dumb, comic book bread is any good. And I’m pretty sure that created a rift in our little community of grape pickers. Anyway, hopefully I pissed of all the Australians and French people reading this — your bread sucks. So, go read the article and have some fun.
