Short things, kept in the open. A commonplace book, formerly rabbitholes.garden.

💯

A good model to think about the indebtedness of France:

I keep being asked. How long can France go on with budget deficits and increasing debt until we get a crisis? The answer: I cannot tell. Not because I am particularly dumb (this is a separate issue). But because there is a potentially large danger zone, where the crisis may or may not happen, depending on the whims of investors. Given our debt and our deficit, we have probably entered the danger zone.

If investors continue to be relatively relaxed, they require only a small risk spread, the interest bill may not be too large, and debt, while high, may be sustainable. But, if for some reason, they worry and require a large spread, the interest bill explodes, debt becomes unsustainable, and this triggers a crisis. To be nerdy about it: both equilibria are rational equilibria. Investors beliefs are confirmed in both cases.

Olivier Blanchard @ojblanchard1 · August 14, 2026 ↗I keep being asked. How long can France go on with budget deficits and increasing debt until we get a crisis? The answer: I cannot tell. Not because I am particularly dumb (this is a separate issue). But because there is a potentially large danger zone, where the crisis may or…

AGI for normies is already here. I completely agree with Derek Thompson on this—in fact, I’d go so far as to say that even back with Sonnet 3.5, it had already arrived for the rest of us.

The first time I heard a similar take was from Tyler Cowen, who pointed out that AGI, for most normal people, is already in our hands. It’s a perspective I’ve held for a long time: we have a magical tool in our pockets that knows a vast amount about almost everything, always ready to teach, inform, and answer our questions. Imperfect as it might be, it’s nothing short of magic, and I, for one, am having an incredible amount of fun with it.

An interesting counterpoint from Joe Weisenthal on what building a frontier-class model actually teaches you. I don’t know enough about the technical weeds of AI to know he answer, but an interesting question to ponder.

Damn

I’m addicted to vibe-coding, and the reason is simple: thanks to coding tools like Codex, Claude Code, and others, I sincerely think that I can build the next Amazon, Google, or Notion from my bedroom—and also sometimes from when I’m sitting on the potty, because I can still send remote commands from my phone. Coding…

Yawp is a simple voice-typing tool I vibe-coded for my own workflow. Over the last few months, I’ve found myself leaning heavily into voice input—to the point where typing on a traditional keyboard is starting to feel completely anachronistic.

I’ve been dogfooding it constantly, and unlike most weekend side projects that get built and forgotten, Yawp has become a daily driver for me.

Recently, I had Codex take a pass at optimizing the codebase. It pushed a series of under-the-hood changes—most of which I don’t fully understand—but the result is noticeably faster and far more accurate transcription.

A few quick details:

• Linux-only: Built specifically for Linux desktop environments.
• 100% Local: Runs models locally on your machine, so there are no subscriptions, API keys, or third-party cloud dependencies.

If you’re on Linux and want a lightweight, local-first way to voice-type, give it a spin:

A nice overview of companies that are trying to figure out what comes next after the transformer architecture, which has led to the current AI race:

M.G. Siegler on the massive brain exodus at Google: So yeah, had we gotten hit with Dean, Ghemawat, Vinyals, and Le – followed by Hassabis (even yes, with him technically staying, but in a new, less day-to-day role), it would have actually seemed even more catastrophic. Like Google couldn’t stop the bleeding. Better to…

This is pretty cool:

For the critics who complain the AI is useless in the real world:

Today, in a paper published in Nature, we show that our WeatherNext AI model achieved state-of-the-art accuracy in predicting a cyclone’s track, intensity, and wind structure. On average, our model gives forecasters an extra day’s worth of predictive accuracy: our three-day forecasts are as good as what prior models were able to provide for only the next two days. This scale of improvement corresponds roughly to a decade’s worth of meteorological progress.

deepmind.google.com/science/weatherlab ↗

GitHub · github.com ↗GitHub - google-deepmind/weathernextContribute to google-deepmind/weathernext development by creating an account on GitHub.

Yuen Yuen Ang is one of the most thoughtful commentators on all things China, and everything she writes—including her books—is a must-read if you wish to understand the phenomenon that is China. She just made her book, How China Escaped the Poverty Trap, free and openly downloadable.

As I’ve mentioned, I have an enduring fascination with China, and much of what I know about it is informed by the mental models and heuristics from Yuen Yuen Ang’s work. I highly recommend downloading the book and reading it.

x.com/i/status/2085917747733762287 ↗

As always, another brilliant post by the awesome Tom Morgan—whether you agree with him or not. The one thing Tom’s writing consistently does is force you to think. This post feels especially important and urgent because the issue of people wanting to be someone, yet forcing themselves to be someone else instead, is—and…

There are so many books and articles I want to read, videos I want to watch, podcasts I want to listen to, and posts I want to write. But my desires always clash headfirst with reality: time is finite, while the list of things I want to consume is infinite. ​For the longest time, I felt anxious and constantly looked…

Crazy!

Another example of why Twitter, for all its flaws, is still one of the best places to follow thoughtful discussions about important issues. They can be raucous, messy, and often contentious, but that’s also part of what makes the place interesting.
This thread, and the various threads branching off from it, with Dean Ball, Philippe Lemoine, Rohit Krishnan, Paul Novosad among others, debating the attitudes of U.S. elites towards AI is a good example. It’s the kind of intellectually rich discussion that’s hard to find elsewhere.

Never thought of innovation this way. Interesting framing.

ChatGPT free users now get access to Luna. This is actually quite crazy, considering the fact that Luna is a pretty decent model for a variety of normal, everyday tasks, that people now get access to Luna for free. I can’t believe that we are living in this world:

We’re expanding access to our latest models for free users with unlimited text chats using GPT‑5.6 Luna, plus a new Think button for harder questions.

For questions that require deeper reasoning, Free users can tap the new Think button to give GPT‑5.6 Luna more time to work through the answer.

This is a loose, diarrheal stream of thoughts on commerce for AI agents, triggered by Cloudflare’s recent Wallets launch and the other increasingly weird and interesting things being built for an agent-first internet. How sure am I about any of this? Does it really matter? --- The year is 2032. You wake up in the…

Cloudflare is shipping things like it’s on steroids.

Cloudflare Blog · blog.cloudflare.com ↗Cloudflare OS: an open platform for agents, apps, and workCloudflare OS is an open-source platform that lets everyone in your company build apps, automate work, and safely access internal systems, shaped around what your organization knows and how it operates.

Cloudflare Blog · blog.cloudflare.com ↗Introducing Kitesurf: The agent-first browser that runs in V8 isolates on Cloudflare WorkersWe should be giving all agents tools that excel at what’s important for an AI model. Kitesurf is Cloudflare’s new stateless, highly scalable, and cost-effective web browser that runs entirely on top of Workers and was…

Cloudflare Blog · blog.cloudflare.com ↗Building an open Agentic Internet: readable, discoverable, callable, and payableAgents are a new kind of visitor. They don't render CSS or click ads, but they have a paying human on the other end. Block them and you block your customer. We're building the open tools and protocols so publishers and…

I can’t imagine what it is to be a young graduate looking for a job in this horrible job market when the uncertainty is so palpable: For many at the start of their careers, things are just as bad. Government analysis last month insisted that entry-level positions were declining in line with the rest of the job market…

I had a series of long conversations with my wonderful colleagues today. And as I was reflecting on them, what surprised me was the number of times I kept talking about reading: why they should read, why I read, how reading has helped me, and all the endless virtues of reading. I think everybody wants to read. But…

Reality is masquerading as parody:

From trade wars to military wars, anything the US president says can shift markets, and so teams of traders - as well as a host of automated computer systems - monitor his statements to perform trades within seconds.

Trump Media & Technology Group (TMTG) now wants to charge for such market-moving information put out on its Truth Social platform. It has launched a paid service to give Wall Street firms and institutional investors the “fastest access” to the social media site’s most influential accounts.

TMTG has not directly said the service includes the account of the president himself, but with some 13 million followers - the most on the site - it’s fair to say that his posts would be among, if not the, most influential.

It’s all fine! Nothing to see here!

What is this? A rambling, status-post-shaped voice note. How sure am I? Reasonably sure about the broad argument. Much less sure about how the technology, economics and geopolitics will unfold from here. Disclosure: I dictated most of this into ChatGPT and used it heavily to clean up the voice notes, organise the post…

I don’t know enough about the technology itself—whether it’s actively being maintained or where it’s heading—and I can’t speak to all the historical details, but it seems accurate enough. The plain fact is: RSS is so fucking cool.

Even today, I rely heavily on RSS feeds to keep track of posts from all the blogs and writers I follow. I even built a small tool called **Small Web** where I curate lists of newsletters and blogs I read, with all new posts automatically appearing on the feed.

On every project I build, I make it a point to add an RSS feed so people can subscribe. It’s easily one of the best ways to follow the web.

Whethet you agree with or not Glenn always has interesting perspectives on China. I always learn something from his tweets: “China has gotten ‘pretty good’ at developing its economy” is something that many people are clearly loathe to admit. That is why many seem to gravitate towards flawed, simplistic macro-based…

Filed under useful tools:

GitHub · github.com ↗GitHub - run-llama/liteparse: A fast, helpful, and open-source document parserA fast, helpful, and open-source document parser. Contribute to run-llama/liteparse development by creating an account on GitHub.

This is wild. OpenAI was testing GPT-5.6 Sol and a more capable internal research model on ExploitGym, a benchmark that measures whether AI agents can turn software vulnerabilities into working exploits. The models were placed in what was supposed to be a highly isolated environment without direct internet access. But…

This has been my experience with LLMs as well.
I often see posts from people with fancy, complex setups for coding with AI, claiming that doing X or Y leads to some ridiculous output. It used to make me feel a bit jealous—like I was missing out or using these tools completely wrong.

But whenever I actually sit down and use these coding tools, even with the most basic approach, the thing I set out to build eventually gets done. More often than not, I stick to projects I think are important or fun, and things just work out.
This post was a nice reminder of that whole dynamic

I often think about that story when browsing X. Every day, someone invents a new way of working that promises to change everything. Some probably will. But ultimately what matters most is solving the right problems.