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

I think this is likely:

Chinese AI developers are increasingly pivoting from general-purpose chatbots towards embedding voice AI assistants into daily applications in search of broader commercial uses for generative AI technologies. The growing industry focus on speech models reflects expectations that voice interfaces could become a key gateway for deploying AI across industries. As one of the most intuitive forms of human-computer interaction, voice requires little user training.

From Kottke:

Calm. Kindness. Kinship. Love. I’ve given up all chance at inner peace. I’ve made my mind a sunless space. I share my dreams with ghosts. I wake up every day to an equation I wrote 15 years ago from which there’s only one conclusion, I’m damned for what I do. My anger, my ego, my unwillingness to yield, my eagerness to fight, they’ve set me on a path from which there is no escape. I yearned to be a savior against injustice without contemplating the cost and by the time I looked down there was no longer any ground beneath my feet. What is my — what is my sacrifice? I’m condemned to use the tools of my enemy to defeat them. I burn my decency for someone else’s future. I burn my life to make a sunrise that I know I’ll never see. And the ego that started this fight will never have a mirror or an audience or the light of gratitude. So what do I sacrifice? EVERYTHING!

I use voice typing heavily because it saves me a lot of time. So I asked Claude/Codex to build me a desktop app. They built a damn good one works like a charm. Still blows my mind that a dummy like me can get this done.

True.

The Bullshit Asymmetry Principle, also known as Brandolini’s Law, states that the amount of energy needed to refute bullshit is an order of magnitude larger than the energy needed to produce it. Coined in 2013 by Italian programmer Alberto Brandolini, it highlights why misinformation and lies spread so quickly and are incredibly difficult to combat

Google summary.

Hat tip to Gappy.

I often wonder how accurate a lot of economic analysis and commentary about India really is, given the fact that India is not really one country. It is many countries in one. Some of our states are as big as other countries. Some of our states are bigger than other countries. Then there is the other problem. When you…

Number 4 is counterintuitive. I always thought that you are not supposed to run ong running sessions. The accepted wisdom is that after 1 or 2 compactions, moving to a new session is a better idea. What Simon here is saying is the exact opposite. I’ll experiment with this

3/ Try using one long-running implementer session for the entire project. Mine routinely run for days or even weeks continuously, compacting many, many times. Compaction works now. Long sessions remember your conventions and patterns and you stop needing to re-explain things.

4/ Drive it with a persistent task list. Your job is to add vetted tasks faster than it can complete them. It’s like shoveling coal into a steam engine. Each task should be: what to do, how to verify, proof notes once completed. The bar is: if satisfied as written, you’d trust the result. Queue as much work as you can before you stop working for the night, especially on Fridays.

 

After launching V4 Flash and V4 Pro, DeepSeek was running a limited-period 75% discount on both models until May 30. Today, it announced that the discount is now permanent. The current pricing is kind of ridiculous. On the official “DeepSeek API pricing page” ( DeepSeek V4 Flash is priced at $0.0028 per million input…

Ken Griffin on AI: Ken Griffin: Let me share a few thoughts on this. First, over the last few months, there has been a step-change function in the productivity of the AI toolkit. It is profoundly more powerful than it was just nine months ago. At Citadel, that has allowed us to unleash a much broader array of use…

So it’s not all hype!

Some of what we describe above is not entirely unique to Mythos Preview. When we ran other frontier models through the same harness, they found a fair number of the same underlying bugs, and in some cases they got further than we expected on the reasoning side too. Where they fell short was at the point of stitching the pieces together. A model would identify an interesting bug, write a thoughtful description of why it mattered, and then stop, leaving the actual chain unfinished and the question of exploitability open. What changed with Mythos Preview is that a model can now take those low-severity bugs (which would traditionally sit invisible in a backlog) and chain them into a single, more severe exploit.

I first read the phrase “garment of destiny” in the wonderful book Fluke, and it’s been stuck in my head ever since. In a real sense all life is inter-related. All men are caught in an inescapable network of mutuality, tied in a single garment of destiny. Whatever affects one directly, affects all indirectly. I can…

I had a few thoughts after watching the movie Train Dreams and instead of writing another cliched douchey review of the movie, I wrote about all the feelings that were swirling in my head as I watched the movie. As I was watching the movie and after I finished watching the movie.

Economic liberalisation reduces emissions and deaths from pollution:

There is a large literature showing that liberalizations boost economic growth. Do these gains come with environmental costs? In this paper, we use 49 cases of large-scale liberalization – as proxied by large discrete jumps on the Economic Freedom of the World Index – to study the effects of liberalizations on greenhouse gas emissions (total, per capita and per dollar of GDP) and deaths from outdoor air pollution. Using matching methods between liberalizers and non-liberalizers, we find that liberalizations have no consistent effect on total emissions or emissions per capita. However, they do reduce emissions per dollar of GDP and reduce mortality from outdoor air pollution. We then employ synthetic control methods on the largest three liberalizers and the largest de-liberalizer and find the same pattern of results.

The AI squeeze is real. While AI may or may not lead to widespread job losses, I think people are underestimating the number of basic job it will kill, jobs that were done by entry level e While AI agents may or may not led to widespread job losses, I’m AI agents will continue to get smarter and more powerful. In my…

Interesting paper from Ken Rogoff and Yuanchen Yang:

Despite these parallels, China is not Japan. The structure of leverage differs: Japan’s crisis centred on private banks and corporations, whereas China’s vulnerabilities are more concentrated in local governments and state-linked entities. China also retains greater administrative capacity to postpone loss recognition and prevent outright financial sector collapse.

China also has other advantages that Japan lacked. Productivity growth remains stronger, and China sits near the global frontier in several fast-growing sectors, including electric vehicles, renewable energy, and more recently, artificial intelligence. These factors may help prevent a full-scale replication of Japan’s stagnation.

But the differences cut both ways. China is ageing faster than Japan did in the 1990s, and as a still-developing country, it lacks Japan’s extensive social safety net. Moreover, newer growth engines, no matter how dynamic, are still small relative to real estate and infrastructure. Rapidly shifting from one export-led boom to another is unlikely to substitute fully for a domestic demand shortfall in an economy of China’s size.

I like this:

We’re socially conditioned to chase what we think everyone else wants. But your true heart’s desire can often be found in the thoughts you gravitate to while undistracted, such as in the shower. As Walt Whitman said, “If you want to know where your heart is, look to where your mind goes when it wanders.”

YouTube · youtu.be ↗What the "Shower Test" Says About Your Life - Jimmy CarrChris Williamson on YouTube

It keeps getting harder to make the case that LLMs aren’t getting better:

The future of Math is mathematicians and AI agents working together.

Very pleased to introduce @GoogleDeepMind’s AI co-mathematician: a multi-agent system designed to actively collaborate with human experts on open-ended research mathematics.

Mathematicians testing the agent across areas as diverse as group theory, Hamiltonian systems, and algebraic combinatorics have reported impressive results.

In autonomous mode evaluation on the rigorous FrontierMath Tier 4 problems, AI co-mathematician scored an unprecedented 48% — a new high score among all AI systems evaluated.

Epoch.

Kevin Rose is building Digg: https://di.gg/ai

I was re-reading Stephen Fry’s Mythos and A.C. Grayling’s Philosophy and Life, and both books reminded me of something I think is still underrated: LLMs as reading companions. Both books are filled with things I don’t know enough about. In the case of Mythos, it’s Greek mythology. In the case of Philosophy and Life…

A nice analysis of long term temperature trends in India.

Amen:

Maybe this sounds banal and corny, but I’ve experienced it as personally profound: you can shift your perspective from “wow, that person is being incredibly stupid” to “maybe they’re currently learning something important,” for any person, at any time. You can also do this to your view of organizations, countries, even the human species.

Most of Your Beliefs About Money Aren’t Actually Yours Money is a made-up thing. Pieces of paper, metal, digital bits that we’ve collectively decided mean something. And yet, despite that, isn’t it funny how large it looms in our lives? How it dictates what we do, how we do it, when we do it? How it colors our moral…

No surprising at all:

Iran’s crypto ecosystem was valued at more than $7.78bn last year, growing at a faster pace compared with 2024, according to crypto transaction monitoring firm Chainalysis. But the data suggests it is not just Iranian citizens who have turned to crypto in a bid to offset the impact of rampant inflation and a weakening currency.

The Islamic Revolutionary Guard Corps (IRGC) accounted for about 50 percent of on-chain activity in the fourth quarter, mirroring its dominance in the country’s economy. Harder to trace and easier to transfer than traditional bank payments, crypto offers a way to sell oil, buy weapons and commodities, circumventing sanctions. And it has also been a method of payment for imports of goods.

Yet Iran’s turn to crypto has also set off a new cat-and-mouse race with the US, with Washington trying to strangle Tehran’s economic options, already limited because of decades of sanctions.

I have a few loose thoughts swimming in my head. Each of them is probably a blog post unto itself, but I’m putting them here so they don’t keep rattling around and escape my brain through the ears. --- Thought One: Attention Reality is far richer than what our feeble brains can comprehend. Even paying momentary…

Therapy speak is probably at the heart of a lot of our problems:

I had a client who described a lot of their problems in therapy language. I told them something like: “hey, I believe that psychotherapy was really useful for you at the time—but consider the possibility that the pathologizing language you picked up from that culture is distancing you from your experience and causing you to consider normal things problematic.” They stopped using therapy framing and found that their inner life got better as a result.

From Nonrequired Reading by Wisława Szymborska: One more comment from the heart: I’m old-fashioned and think that reading books is the most glorious pastime that humankind has yet devised. Homo Ludens dances, sings, produces meaningful gestures, strikes poses, dresses up, revels, and performs elaborate rituals. I don’t…

Makes you think:

One thing I find fascinating about LLMs is that they reveal how much of the quality of our relationships comes down to our capacity to be open and prompt others in a good way. Here we have, for the first time, millions of people having conversations with the precise same entity (Claude, ChatGPT), and they are having wildly different experiences—some hate it, find it stupid and boring, others get it to say and do fantastic things.

This is true of interacting with people, too (though of course I’m not saying interacting with an LLM is anything like interacting with a person)—the quality of a relationship has a whole lot to do with your skill in prompting/asking the right question/providing good context and space for the other. But previously it was tempting to assume that differences people experience had more to do with some people being lucky in getting to interact with more interesting people vs others only had met boring people. Though that of course is also a factor.

Addy Osmani on harness engineering:

A decent model with a great harness beats a great model with a bad harness. I’ve watched this play out on my own work over and over. And increasingly the interesting engineering isn’t in picking the model, it’s in designing the scaffolding around it.

This is the first open-weight model that plays like a top-class agentic model. Watching it go through ambiguous and meticulous chained tool work successfully puts it squarely in the wheelhouse of Opus 4.6. We’re looking at an open weight model, but with much cheaper direct inference provider pricing. For a subclass of our eval set, it’s outperforming GPT 5.2. We’re about to undergo a gigantic industry shift.

Man has it all in his hands, and it all slips through his fingers from sheer cowardice.

Fyodor Dostoyevsky, Crime and Punishment

Post type: Stream of consciousness. How sure am I? Reasonably sure, until the next model upgrade. Note: I dumped these as voice notes into Claude and had it clean them up. I wouldn’t want to be a young graduate or postgraduate looking for a job right now. If my LinkedIn inbox is any indication of the state of India’s…