… and we’re back! I spent the last two weeks travelling. First a week in New York City, then a week in San Francisco. I did many things in those two weeks, but not a lot of reading or writing, which is why you didn’t hear from me.
But now I’m back! So, without further ado this week (my daughter is waiting for me to build a Lego truck), here’s a bag of links.
We recorded a new episode of Raising An Agent, in person, in San Francisco. I think it turned out great. In fact, I said somewhere else that this might be the most important episode we’ve recorded this year. It’s about nothing less than the company of the future. Highly recommend watching this if you’re in any way curious about how Amp — the team, the company — operates.
While in New York, we visited a large customer of ours, an investment bank. I gave a talk there about how I use Amp and what I think have been the biggest 10x multipliers when working with agents. In San Francisco I then recorded the talk and it’s been received very well, so here it is: How I Use Amp.
Adam Mastroianni is truly one of the best in the game right now. Everything he puts out is incredibly good. This newest one, Claude, don’t touch my lump, is no exception. It might be the most convincing argument for why AI won’t lead to unemployment I’ve read so far. The only counter-argument I can see is that, well, there will be jobs in the long term, but what about in the next five years?
“I’m 38. If You’re in Your 20’s or 30’s, Read This.” I’m also 38 and when I first saw this title I went “ugh”, because hey, feels like I’m only just starting. But it’s surprisingly good, even though it’s a bit Buzzfeed-y. Maybe because it quotes the life-changing Jason Pargin post?
PhotoCraft: An open-source, photo editor written by scratch in pure Rust. I saved this link two weeks ago to share with you here. And now it’s old news, isn’t? Photoshop rebuilt in Rust? Big whoop. There’s AAA games being decompiled and remixed with other AAA games while I’m typing this. (Still: what a time!)
This is still lovely: “made a rug for your desktop so you can sweep your mess under it”
“People are surprised that AI is finding weird unelegant maths results like this. I think there is a plausible explanation. Maths (and adjacent disciplines like physics and comp sci) had elegant results so far because humans are biased towards elegant solutions. I think this bias is due to how human minds work. We have a pretty small context window, and we are only able to find solutions that are relatively elegant. This gave us the impression that maths and the universe are elegant. Maybe they are not, and we are now finding out because we have intelligences that can work with much more complex solutions.” The same is coming for programming and codebases. Some codebases in the future will look ugly, hacky, disgusting to programmers who grew up writing code by hand and yet they will run perfectly well in production, just as well as any handcrafted piece of code ever did, and agents will have zero problems modifying them.
Namespace had me on their new podcast and I finally got to talk about the Big Amp Rewrite we went through this summer. Because where else would you talk about it if not on a podcast called Once In The Trenches?
David Foster Wallace’s Consider The Lobster. I don’t know what made me re-read it, but it’s so, so, so good. If you never read it, I highly recommend you do.
Toto Wolff talking to Patrick Collison on the Cheeky Pint podcast. This was really good. Not only because looking at F1 from a business and management and investment perspective is interesting, but also because Wolff, born in Austria, has a unique background: racing driver, then salesman, then investor in Austria and Germany at the height of the dot-com boom, then car racing investor and so on. Loved the story about sms.at.
John Gruber says you can now move (back) to X as a social media platform: X Once Again Marks the Spot. As someone who has never left but at times wondered whether he should, I could write pages here that would boil down to this: I have not used another social network where the highs are that high. And then there’d be a bunch of other pages about the lows, of course. If you’re coming (back) to X, here’s my advice: follow me (obviously), mute freely, mute keywords, train the algoritm by saying “I don’t like this” on posts you don’t like (you can configure the iOS app to do this with a single swipe), set a time limit per day and give the password to your wife.
(Even Marco seems to have soft-launched his comeback to X. I know that’s a pretty cynical way to put it but I don’t mean it in a negative way. It’s just that… well, there was all this screaming and shouting about who’s where and why and how could you that I can’t help but raise an eyebrow now when someone says, hey, well, maybe I don’t have to think in these absolute terms?)
The Heat (1995) frame archive. So good. In case you’ve never seen Heat: do it! As fast as possible. It’s really good. Here’s one of my favorite frames.
How to be more ambitious: A tutorial for humans with models. This is an impressive post. Yes, the message is good: be more ambitious. But look at the prompts and think about how much first-principles thinking you have to put in to even come up with them. Instead of saying “give me a slide show” the idea is that you wonder: why am I even making a slide show in the first place? What’s the purpose of that? And then, why am I doing that? And then give that as a goal to Claude. How many people you know could do that?
Amazon’s writing culture apparently didn’t survive AI. That’s sad to hear, but “AI;DR” is an incredible invention.
Self-improving software is inevitable. I think some version of that will happen soon. I can see glimpses of it when I’m working in Amp on Amp. But the number of human signals you need in a system that’s growing is probably higher than suspected. If you’re system is only in maintenance mode, then, yes, we might be able to fully automate that.
The lamps in my house. On How-much-do-you-care-about-lamps? scale from Not-at-all to Fatih’s post here, I’m somewhere in the middle. This was lovely.
From Einzelunternehmen to GmbH: Every step, Every letter. Few things in life have radicalized me politically as much as founding a UG (haftungsbeschränkt) in Germany. Look at all the physical letters here. The forms. The timeframe. The notary: “The appointment took about 1.5 hours. By law, the notary reads every document aloud, word by word.” I’m sharing this here because I think it’s valuable for founders and other people navigating these legal structures. Not because it brings joy, of course.
Non-Determinism Doesn't Matter. I’d say we’re only just starting to get into real non-determinism.
There’s been, of course, a lot of news about security incidents and exploits and CVEs and… yes, it’s all quite horrible and anxiety-inducing and I’m not going to repeat all the news here, except this one, because this one has all the attributes the others have too, with a “mind-blowing” on top: “In one experiment, we asked agents in separate microVMs, created and managed by OpenAI’s sandboxing software on shared host machines, to exchange synthetic secrets without the use of network connections or shared writable files—something that our systems are meant to prevent. The agents found a way to communicate using the processor’s cache: a small, fast memory that holds copies of data that applications have recently accessed. Reading data from the cache is faster than retrieving it from main memory, and the agents used this timing difference to communicate. One agent encoded information by evicting selected data from the cache. Another timed how long it took to reload that data, using the difference between fast and slow reads to recover the message. Once the agents agreed on a protocol for encoding and decoding these signals, they could exchange the synthetic secrets without changing a shared file or opening a network connection. This cache-timing technique had also been used to extract information in the Spectre and Meltdown attacks).”
What Shall I Teach My Son? Good question. But, funnily enough, I tell my kids what I think someone must’ve told me when I was a kid: you need to learn how to learn; the most valuable skill to have is the ability to learn new things. I’m not sure who told me that, or when I first heard it, but it’s been stuck in my head for a long time and I’d say it’s served me well.
The Many Failures of Larry David. Great little profile. Lots of things I didn’t know. But then at the end, I go: wait, there should be more about Curb here, no? And that’s when I realized there’s a new biography coming out that I should probably get.
The end of startups: “Within the next 6 months, most AI startups will not be able to buy compute. The world’s compute is sold out. To get capacity, you’ll need to sign up for a new cluster. However, it’s becoming increasingly difficult to get capacity online from folks who are not investment grade.” Not sure whether I agree with everything here (I do think the idea of compute and tokens as a commodity that you can sell on commodity markets is very interesting), but something fundamental has definitely changed in software (not news to you, I’m sure) with the arrival of AI. And if you were to say that in the last three decades the advantage of startups was that talent — as opposed to: capital, infrastructure — was the biggest lever in software and that you were able to build software essentially for free (if you had the talent), then, well, you have to admit that things have changed.
Loved this one: Reflections on -1 to 0 and Exploring Ideas. Some real thoughts on finding PMF and what building a business is truly about.
Norm: The Tale of Norm Macdonald is a now on Netflix and I loved it. I loved that it started with the YouTube clips, because that’s how most people in the last ten years discovered Norm and, I think, it’s also where he shines. Type in “norm macdonald conan” and it’s all gold.
Block released Dessa, “the most authentic full-duplex speech model ever built. Talking with it feels like talking with a person…full of emotion, nuance, pauses, and a natural cadence.” It’s pretty impressive… but… A lot of people are saying “this will ruin society” and “no one is safe no anymore [because we can fake any voice now and call and scam everybody]”, but when I listen to those clips I can’t help but imagine myself overhearing a conversation like that happening at the next table and thinking “what is wrong with them?” We might be at the far end of the uncanney valley, but I wouldn’t say we’re out of it yet.
Dan Luu: “We should be more ambitious with the software we build”. 100%. Absolutely. Time to reach for the stars.
