Rabbit Holes

Google Gemini 3 API Is better than I expected

Bhuvan — Fri, 03 Apr 2026 00:00:00 GMT

Gemini’s web chat app is unusable and, to be honest, despite all the hype and hoopla, it kind of sucks. But surprisingly, my experience using their APIs has been really, really good, even compared to the likes of Claude and OpenAI in a lot of cases.

For a lot of these high-volume tasks, I’ve been heavily using Gemini 3 Flash Preview for Project Akshara to digitize books at scale, and Flash has been phenomenally reliable. It has turned out to be a wonderful workhorse model.

For the longest time, I could never really make Gemini 3 and 3.1 Pro work because of their surprisingly low rate limits. But I was trying them on a few translation tasks, and again, they were surprisingly much, much better. In one particular case, I was working on modernizing Francis Bacon’s essays, and it did much better than Sonnet. And I’m saying this as a huge fan of Claude.

So, if you’re looking for a reliable workhorse model for high-volume tasks like processing information, cleaning it up, organizing it, extracting text from PDFs and images, and so on, my experience with Gemini 3 Flash Preview has been really, really good. Oh, and it helps that Gemini Flash 3 is dirt cheap.

On whether Russia can continue to recruit new soldiers.

Bhuvan — Wed, 01 Apr 2026 00:00:00 GMT

Russia is not running out of new soldiers to send to the meat grinders in Ukraine:

The main sources of data on new recruits signing up are federal and regional budget spending reports, which include a separate section on bonuses paid for signing military contracts. The sizes of the bonuses paid at both the federal and most regional levels are a matter of record, which makes it possible to estimate the numbers of new military recruits on at least a quarterly basis. Based on these data, over the past two years, the Russian armed forces and associated private military companies have recruited 30,000–40,000 contract soldiers every month, or 1,000–1,300 people per day.

Meanwhile, recruitment is becoming increasingly expensive. The federal authorities doubled their payments back in 2024, while the regions, which are set recruitment targets by Moscow, drastically increase signing-up bonuses during periods of candidate shortages and reduce them again once quotas have been met. While the federal budget can still afford to do this, these expenses are becoming a significant burden for the regions.

Regional budgets are increasingly in debt, but their collapse is not imminent. The regions are cutting spending on other, “lower priority” items. The central government is providing top-ups from the federal budget to offset some of these costs.

Feeling god up

Bhuvan — Sun, 29 Mar 2026 00:00:00 GMT

I'm not a religious person. I think there are traces, but more or less, I'm not a religious person. I have few traces remaining, thanks to the fact that I grew up in a religious household, and my parents are still deeply religious. Depending on the day you ask me, and depending on what I've read the previous day, I might call myself spiritual, atheist, hard or soft determinist. But religious I am not.

That said, I think one of the closest instances I've come to seeing God is when reading the sublime verses in a poem. Tagore, Dickinson, Mary Oliver, William Blake. You can't help but take a shovel and pitchfork to the hardened layer of rationalist sediment on your psyche, uproot it all, and make way for some other feeling, some non-cold feeling of vibrancy. Though with poetry, I think I'm more inclined towards what Baruch Spinoza had to say: that God and Nature are one and the same thing.

And when you lie down on your back on a beautiful warm summer night, and gaze up at the heavens to see fluffy white clouds parting to reveal a crescent moon and distant stars, you know you are in the presence of something truly grand. I don't know if that is an all-knowing deity, wise and powerful, magic wand in hand, stirring things up. But this, to me, is also feeling the hand of God.

I resonate with what Elizabeth Bruenig is saying:

To imagine that one might find traces of the divine strewn throughout the universe, or that earthly methods of inquiry might uncover some of those signs, isn’t ridiculous. But this latest round of arguments in favor of intelligent design seems aimed mostly at establishing that God could or should exist within the rational frameworks we already employ. This is both weak grounds for belief and a fundamental misunderstanding of faith. The route to durable faith in God often runs not through logical proofs or the sciences, but through awe, wonder, and an attunement to the beauty and poetry of the world, natural and otherwise.
This was not always apparent to me.

I came to this understanding through trial, error, and my own brushes with scientific rebuttals to the existence of God.

Here's what God sounds like:

To see a World in a Grain of Sand
And a Heaven in a Wild Flower
Hold Infinity in the palm of your hand
And Eternity in an hour
A Robin Red breast in a Cage
Puts all Heaven in a Rage
A Dove house filld with Doves and Pigeons
Shudders Hell thro all its regions
A Dog starvd at his Masters Gate
Predicts the ruin of the State
A Horse misused upon the Road
Calls to Heaven for Human blood
Each outcry of the hunted Hare
A fibre from the Brain does tear
A Skylark wounded in the wing

I got some links for you

Bhuvan — Sun, 29 Mar 2026 00:00:00 GMT

This week's digest

Venkatesh Rao is consistently one of the most thoughtful people online. He is one of my favorite writers and one of the sharpest observers of whatever is happening around us, especially in technology. I have been following his thoughts on AI, specifically on making things with AI and more specifically on writing with AI.

Right now, there is a lot of angst around this. A lot of it, frankly, is just stupidity: this idea that if writing is touched by AI, it immediately stops being good. “AI-written” or “AI-assisted” has become a kind of weird epithet, as though all human writing is some sanctified thing that exists in a divine, unsullied realm. And somehow, the moment AI interacts with it, it gets tainted and becomes bad.

There is clearly a bit of a moral panic going on. People are incessantly labeling anything AI touches as a flop, or as inherently inferior, and so on. It is not only irritating, but also funny, because people seem to forget something basic: it is not AI that is flawed. Most people are flawed. People just need the sense to listen to themselves speak for a while, maybe in front of a mirror, and they will realize that even the median stochastic vomit produced by an LLM is better than 80 percent of the things we humans say.

But anyway, this is a weird moment. A new technology has entered the fray and is reshaping the way people think, create, write, and so on. This is a huge technological shift, and these old-fashioned, puritanical opinions are probably part of the course.

Take readers’ opinions seriously, but not too seriously. I’d say 1/6 have real spiritual-angst resistance. Ignore them. Another 1/6 will never get past attachment to gleegloom over the not-X-but-Y “tell” du jour. Ignore them too. Counterintuitively, at the other end, 1/3 of readers will be delusionally committed to declaring any and all AI writing “good” (similar to how committed feminists might insist that any work done by women is necessarily good). Ignore them too. The middle third is the target audience. People who will simply treat AI writing like human writing — keep reading if they like a particular piece but abandon if they don’t. You want those who discriminate at piece level, not class level (AI classists/racists?).

Finally: Reading good AI generated writing is a new kind of literacy that’s currently evolving from pidgin to creole. Most people not only lack this literacy (beyond having rudimentary hostile radars; a bit like being on alert for human accents that trigger distrust for you), they are unaware there even is a literacy taking shape. Developing this literacy yourself is the main goal for you as a writer. Catalyzing it in the eligible subset of readers is a nice extra. For me, this practically translates to a personal re-read test. Do I go back and read one of my own sloptraptions for the content later? About a third or so pass this test strongly.

Get ready to be a part of the permanent underclass!

While the laborers who were displaced by machines during the industrial revolution were able to move into new and growing service industries, allowing the economy to keep expanding and unemployment to stay low while incomes and living standards rose, with the higher productivity that comes from technological progress, there are major doubts about where the jobs and incomes will come from if AI is eventually able to do everything that workers do today. These same doubts have been around ever since the Luddites, yet the progress that follows rising productivity and new technologies has always created new industries that provide incomes for labor.

However, one of the consequences of digital technology has been an increasing share of income going to the owners of capital and a declining share to labor (Exhibit 1A). AI is likely to exacerbate this trend even more. The growing share of capital as opposed to labor income is also implicit in the rising level of wealth relative to income. Since the 1960s the ratio of household wealth to income has risen from around five times to about eight times (Exhibit 1B).

An important implication of this transition from most labor income to mostly capital income is a need to broaden ownership of capital to mitigate the effects of a shrinking labor income share on the broader population. To some extent, this is already happening, at least in the U.S., where there is much wider ownership of Equities compared to most other economies. Without policies to make the ownership of capital and its growing income share more widespread, political polarization and instability are likely to increase as we have seen in recent years.

This has been my experience as well. I am now doing more things than ever, both personally and professionally. In fact, after LLMs, I'm starting to feel like there isn't enough time to do all the things I want to do. Work fills available tokens?

It's remarkable how much of my work is completely automated w/ AI, and yet, I still am necessary. The amount of time I personally have to spend working just isn't going down. Instead, the leverage of my own time is going up. Every second I spend not working becomes more painful

I have been working less and less recently. I still work a crazy amount. More than I ever have. Every spare second. It's just that, as my family grows, the opportunity cost of not working equally goes up. That's the only thing that is saving me, I think

Tweet link

I also relate to the family angle. Given the miserable state of the world, my feeling that we're headed for an age of discord and increased economic instability only continues to become stronger. Which means LLMs now give me the leverage to work harder.

China is the new pharma superpower:

Until recently, China’s drug industry was best known for making generic drugs, supplying ingredients and running trials for Western firms. Over the past decade it has reinvented itself. Approval processes have been streamlined, priority reviews conducted for drugs tackling critical conditions, and regulations brought closer to international standards. The workforce at China’s drug regulator quadrupled between 2015 and 2018, and a backlog of 20,000 new drug applications was cleared in just two years. The time taken to secure approval for human trials shrank from 501 days to 87. And the output of new medicines soared. In 2015 China approved only 11 treatments, mostly Western imports. By 2024 the figure had risen to 93, with 42% developed domestically.

The illusion of technological omnipotence:

Yet never has so much been seen, so precisely, by so many people who understand so little of what they are seeing. A system can tell you where a man is. It cannot tell you what his death will mean for a nation. Such systems are trained on behavior, not on meaning — they can track what an adversary does but not what he fears, honors, remembers or would die for.

This is the recurring illusion of overequipped leaders: Because they can map the battle space, they think they understand the war. But war is never merely a technical contest. It is shaped by grievance, sacred narrative, the memory of past humiliations and the desire for revenge. Those are not atmospheric complications added to an otherwise technical enterprise. They are what the war is about.

The pain of living with pain

Bhuvan — Sat, 28 Mar 2026 00:00:00 GMT

This is a beautiful post by Ted Gioia on the mysterious pain that afflicted him and how we went about figuring out the exact cause.

The first time this happened, I jumped out of bed and ran to get an ice pack from the freezer. I pressed the ice pack to my face and—surprise!—this just made the pain even worse. I’m fortunate that the fire only lasted around sixty seconds, but if I tried to go back to sleep, it could happen again a few minutes later.

My nights now became perilous. I found that the only way to sleep without risk was to remain absolutely motionless on my back—even the slightest movement could bring on another attack.

Revenge of the real

Bhuvan — Fri, 27 Mar 2026 00:00:00 GMT

Absolutely

The second lesson is that we are in an age when those “hard industries” matter. Yes, Big Tech stocks have soared in recent years amid excitement about AI. But the Iran war shows that countries are very vulnerable if they lack industrial processes, however old-fashioned. “The old economy begins to take its revenge,” Jeff Currie, analyst at the US private capital group Carlyle, recently noted. “You can’t print molecules,” even with AI.

This time is NOT different

Bhuvan — Thu, 26 Mar 2026 00:00:00 GMT

Nick Maggiulli writes that AI disruption will be the same as previous technological shifts:

Some will argue that “this time is different” because AI is replacing knowledge work, not just physical work. And, if this comes to pass, what will there be left for people to do?

It’s a compelling argument, but people made similar arguments about the automation of physical work. There was literally a group of people in England in the early 1800s called the Luddites who destroyed weaving machines due to their adverse impact on textile workers.

The Luddites couldn’t imagine what would replace their livelihoods and the same is true for us now. There will be future roles that require a different set of human skills that we can’t even imagine today. These skills won’t directly compete with LLMs, but will enhance them.

The internet gets shittier and shittier

Bhuvan — Sun, 22 Mar 2026 00:00:00 GMT

This is sad:

The Times says the move is driven by concerns about AI companies scraping news content. Publishers seek control over how their work is used, and several—including the Times—are now suing AI companies over whether training models on copyrighted material violates the law. There’s a strong case that such training is fair use.

Whatever the outcome of those lawsuits, blocking nonprofit archivists is the wrong response. Organizations like the Internet Archive are not building commercial AI systems. They are preserving a record of our history. Turning off that preservation in an effort to control AI access could essentially torch decades of historical documentation over a fight that libraries like the Archive didn’t start, and didn’t ask for.

If publishers shut the Archive out, they aren’t just limiting bots. They’re erasing the historical record.

Time to work hard

Bhuvan — Sat, 21 Mar 2026 00:00:00 GMT

Good advice from Tyler Cowen:

If strong AI will lower the value of your human capital, your current wage is relatively high compared to your future wage. That is an argument for working harder now, at least if your current and pending pay can rise with greater effort (not true for all jobs).

Invictus by William Ernest Henley

Bhuvan — Fri, 20 Mar 2026 00:00:00 GMT

Beautiful 🤩

Out of the night that covers me
Black as the pit from pole to pole,
I thank whatever gods may be
For my unconquerable soul.

In the fell clutch of circumstance,
I have not winced nor cried aloud.
Under the bludgeonings of chance
My head is bloody, but unbowed.

Beyond this place of wrath and tears
Looms but the Horror of the shade,
And yet the menace of the years
Finds, and shall find, me unafraid.

It matters not how strait the gate,
How charged with punishments the scroll,
I am the master of my fate
I am the captain of my soul.

Today's links

Bhuvan — Wed, 18 Mar 2026 00:00:00 GMT

Jensen in the Arena Om Malik on OpenAI's IPO plan Where OpenAI’s technology could show up in Iran

kottke.org turn 28

Bhuvan — Sun, 15 Mar 2026 00:00:00 GMT

One of my favourite sites on the internet. I hope it survives the end of time itself:

28 years of kottke.org, as of today. Older than Google. Older than The Matrix. Older than Christopher Nolan’s feature film career. Older than Elle Fanning. Older than Kurt Cobain when he died. 47,300 posts since March 14, 1998. It might outlast American democracy.

Found the link on https://smallweb.blog

Attention is first philosophy

Bhuvan — Sun, 15 Mar 2026 00:00:00 GMT

I'm a big fan of Adam Robbert and love his framing of attention as the first philosophy. A few snippets from his recent talk:

1. Attention is the practice beneath all philosophical practices. Nearly every contemplative tradition—Stoic, Platonic, Christian monastic—was built on attention as the foundational practice. Practices like fasting, physical training, and meditation exist to guard the stillness attention requires. Philosophy, in this sense, is a discipline of perception, not mere argumentation.

2. Distraction is a perennial human problem, not a modern pathology. Medieval monks complained that illuminated manuscripts were too decorative and distracting from the text. Socrates bemoaned writing as a technology that would erode our memory. But now, we accept writing as part of the intellectual life. The shape of the distraction changes, but the underlying dynamic doesn’t.

3. Human memory is an ecosystem that must be tended to. Human memory doesn’t store files like a computer, but reorganizes perception at the level of physical sensation. What you attend to shapes your memory, and how reality shows up to you. Writer Eleanor Robins offers a metaphor: your memory is like a garden that you must fertilize with rich material; overconsumption of flat, repetitive media leads to the “desertification” of the inner life.

Craig Mod on building software using Claude

Bhuvan — Sat, 14 Mar 2026 00:00:00 GMT

You can just do things! Like building your own accountant!

Simply put: It’s a big mess, and no off-the-shelf accounting software does what I need. So after years of pain, I finally sat down last week and started to build my own. It took me about five days. I am now using the best piece of accounting software I’ve ever used. It’s blazing fast. Entirely local. Handles multiple currencies and pulls daily (historical) conversion rates. It’s able to ingest any CSV I throw at it and represent it in my dashboard as needed. It knows US and Japan tax requirements, and formats my expenses and medical bills appropriately for my accountants. I feed it past returns to learn from. I dump 1099s and K1s and PDFs from hospitals into it, and it categorizes and organizes and packages them all as needed. It reconciles international wire transfers, taking into account small variations in FX rates and time for the transfers to complete. It learns as I categorize expenses and categorizes automatically going forward. It’s easy to do spot checks on data. If I find an anomaly, I can talk directly to Claude and have us brainstorm a batched solution, often saving me from having to manually modify hundreds of entries. And often resulting in a new, small, feature tweak. The software feels organic and pliable in a form perfectly shaped to my hand, able to conform to any hunk of data I throw at it. It feels like bushwhacking with a lightsaber.

Useful markdown tools

Bhuvan — Sat, 14 Mar 2026 00:00:00 GMT

Defuddle and markdown.new.

Good writing advice from Frederik Gieschen

Bhuvan — Sat, 14 Mar 2026 00:00:00 GMT

A writing maxim I want to tattoo on my hand: Stop trying be smart. Just be honest about how you feel.

Who pays the price for the war on Iran?

Bhuvan — Sat, 14 Mar 2026 00:00:00 GMT

Depressing chart by Rafael.

Why the Strait of Hormuz is important

Bhuvan — Thu, 12 Mar 2026 00:00:00 GMT

Why the Strait of Hormuz is important.

All the advice you need in life. Blake's Proverbs of Hell

Bhuvan — Wed, 11 Mar 2026 00:00:00 GMT

Here is all the advice you’ll ever need in life, courtesy of the great William Blake. Throw those fucking garbage self-help books away. Actually, don't just throw them—give them a loving send-off by using the pages as toilet paper to wipe your ass. It’s a much better use for them.

Proverbs of Hell

From The Marriage of Heaven and Hell — William Blake (Project Gutenberg)

In seed-time learn, in harvest teach, in winter enjoy.

Drive your cart and your plough over the bones of the dead.

The road of excess leads to the palace of wisdom.

Prudence is a rich ugly old maid courted by Incapacity.

He who desires, but acts not, breeds pestilence.

The cut worm forgives the plough.

Dip him in the river who loves water.

A fool sees not the same tree that a wise man sees.

He whose face gives no light shall never become a star.

Eternity is in love with the productions of time.

The busy bee has no time for sorrow.

The hours of folly are measured by the clock, but of wisdom no clock can measure.

All wholesome food is caught without a net or a trap.

Bring out number, weight, and measure in a year of dearth.

No bird soars too high if he soars with his own wings.

A dead body revenges not injuries.

The most sublime act is to set another before you.

If the fool would persist in his folly he would become wise.

Folly is the cloak of knavery.

Shame is Pride’s cloak.

Prisons are built with stones of law, brothels with bricks of religion.

The pride of the peacock is the glory of God.

The lust of the goat is the bounty of God.

The wrath of the lion is the wisdom of God.

The nakedness of woman is the work of God.

Excess of sorrow laughs, excess of joy weeps.

The roaring of lions, the howling of wolves, the raging of the stormy sea, and the destructive sword, are portions of Eternity too great for the eye of man.

The fox condemns the trap, not himself.

Joys impregnate, sorrows bring forth.

Let man wear the fell of the lion, woman the fleece of the sheep.

The bird a nest, the spider a web, man friendship.

The selfish smiling fool and the sullen frowning fool shall be both thought wise that they may be a rod.

What is now proved was once only imagined.

The rat, the mouse, the fox, the rabbit watch the roots; the lion, the tiger, the horse, the elephant watch the fruits.

The cistern contains, the fountain overflows.

One thought fills immensity.

Always be ready to speak your mind, and a base man will avoid you.

Everything possible to be believed is an image of truth.

The eagle never lost so much time as when he submitted to learn of the crow.

The fox provides for himself, but God provides for the lion.

Think in the morning, act in the noon, eat in the evening, sleep in the night.

He who has suffered you to impose on him knows you.

As the plough follows words, so God rewards prayers.

The tigers of wrath are wiser than the horses of instruction.

Expect poison from the standing water.

You never know what is enough unless you know what is more than enough.

Listen to the fool’s reproach; it is a kingly title.

The eyes of fire, the nostrils of air, the mouth of water, the beard of earth.

The weak in courage is strong in cunning.

The apple tree never asks the beech how he shall grow, nor the lion the horse how he shall take his prey.

The thankful receiver bears a plentiful harvest.

If others had not been foolish we should have been so.

The soul of sweet delight can never be defiled.

When thou seest an eagle, thou seest a portion of Genius. Lift up thy head!

As the caterpillar chooses the fairest leaves to lay her eggs on, so the priest lays his curse on the fairest joys.

To create a little flower is the labour of ages.

Damn braces; bless relaxes.

The best wine is the oldest, the best water the newest.

Prayers plough not; praises reap not; joys laugh not; sorrows weep not.

The head Sublime, the heart Pathos, the genitals Beauty, the hands and feet Proportion.

As the air to a bird, or the sea to a fish, so is contempt to the contemptible.

The crow wished everything was black; the owl that everything was white.

Exuberance is Beauty.

If the lion was advised by the fox, he would be cunning.

Improvement makes straight roads, but the crooked roads without Improvement are roads of Genius.

Sooner murder an infant in its cradle than nurse unacted desires.

Where man is not, nature is barren.

Truth can never be told so as to be understood and not to be believed.

Enough! or Too much.

Robin Williams on intelligent design

Bhuvan — Tue, 10 Mar 2026 00:00:00 GMT

This is hilarious 😂

https://youtu.be/we262Cp4eUE?si=S82LsTaYEDKXpwXq

Gangster imperialism with a side of dick measuring contest

Bhuvan — Mon, 09 Mar 2026 00:00:00 GMT

Brilliant piece by Eskandar Sadeghi-Boroujerdi:

In a farcical re-enactment of the Iraq War script, we were told that the Islamic Republic would collapse like a house of cards. But unlike in 2003, there has been little attempt to persuade the wider world, or even the US Congress. The rhetorical labour that accompanied the invasion of Iraq, however flawed or dishonest, has largely been abandoned. Even senior US military officials have struggled to explain how the campaign’s objectives would be achieved swiftly or decisively. The assumption of inevitability has replaced the burden of argument.

The absence of justification is not incidental. It is a morbid symptom of an international system in crisis. The certainties of the United States’ hegemonic stewardship of the ‘international rules-based order’ have been deformed beyond recognition by the Gaza genocide, but no alternative architecture has cohered in their place. Instead there is a politics of gangster imperialism that has neither international nor domestic consent.

Violent spectacles

Bhuvan — Sun, 08 Mar 2026 00:00:00 GMT

Brilliant editorial in the equator magazine:

The attempt to enforce American primacy by military means led to catastrophe in Helmand and Fallujah, just as Wall Street imploded at home. What has emerged from that failure is something more dangerous than either the Cold War order or the unipolar moment. The US may still be hegemonic in Europe, where obedient satraps like Merz and von der Layen genuflect to Washington, but for the rest of the world, America has stopped being either leader or lodestar. “The unravelling of the neoconservative project,” Giovanni Arrighi wrote in 2009, “has for all practical purposes resulted in the terminal crisis of US hegemony – that is, in its transformation into mere domination”.

Under the current regime, the US has little left to offer the world but a shameless display of coercion and destruction. Trump and his lieutenants seem intoxicated by their own impunity, indifferent to international law, uninterested in manufacturing consent, and exercising a form of political gangsterism through intimidation, kidnapping or offing rival heads of state. “Force,” Weil wrote, “is as pitiless to the man who possesses it, or thinks he does, as it is to its victims. The second it crushes; the first it intoxicates.” The stupefaction that descends on those who discover they can act without consequence is what she saw as arguably the deeper and more enduring catastrophe. The wielder of force experiences their own power as the natural order of things, and cannot imagine that it could be otherwise, or that the people their force is directed against – women, immigrants, Muslims, progressives and liberals – are human beings at all.

Harrison Ford: Life Achievement Award Acceptance Speech

Bhuvan — Sun, 08 Mar 2026 00:00:00 GMT

Beautiful:

As actors, we get to live many lives. We get to explore ideas that affirm and elevate our shared experience. The stories we tell have a unique capacity to create moments of emotional connection. They bring us together. So while we're all at different stages of our lives and careers in this room, we all share something fundamental. We share the privilege of working in the world of ideas, of empathy, of imagination. Sometimes we make entertainment. Sometimes we make art. Sometimes if we're lucky we make them both at the same time. And if we're really fortunate, we also get to make a living doing it.

https://youtu.be/dV_2CEa6Bbs?si=D65RNnIsQ6mBVqmW

Evaluating the success of China's solar industrial policy

Bhuvan — Sat, 07 Mar 2026 00:00:00 GMT

Interesting paper on the success of China's solar industry:

China’s solar success gives hope for industrial policy, but it cannot necessarily be replicated in all sectors. Part of the success is likely because solar is a relatively young, high-tech industry, with more opportunities for learning than a more mature industry such as shipbuilding (Kalouptsidi 2018).

These results tell us that China's solar subsidies worked for Chinese citizens. What they did for the world remains an open question. China may have squeezed out solar industries elsewhere, or the knowledge it generated may have spilled across borders fostering learning elsewhere just as it spilled over from city-to-city within China. What happened to global innovation, production and decarbonisation is the subject of ongoing research.

Whatever China's industrial policy did to solar industries elsewhere, its most tangible legacy is solar panels cheap enough that the world is actually using them – not out of environmental conscience, but because they are the cheapest option. China's taxpayers funded an investment that paid for itself, and this in turn cheapened solar as an energy source and reduced emissions.

Tamil inscriptions on the tombs of Egyptian kings

Bhuvan — Sat, 07 Mar 2026 00:00:00 GMT

This is so cool. Goes to show the extent of ancient trade links.

A path-breaking finding has shed new light on trade links between ancient Tamilagam, other parts of India and the Roman Empire. Two researchers have identified close to 30 inscriptions in Tamil Brahmi, Prakrit and Sanskrit at tombs in the Valley of the Kings in Egypt. These inscriptions are said to belong to the period between the 1st and 3rd Centuries C.E.

These inscriptions were identified during a study carried out in 2024 and 2025 by Charlotte Schmid, Professor at the French School of Asian Studies (EFEO) in Paris and Ingo Strauch, Professor at the University of Lausanne in Switzerland. The team documented them across six tombs in the Theban Necropolis. They followed the footsteps of French scholar Jules Baillet, who surveyed the Valley of the Kings in 1926 and published more than 2,000 Greek graffiti marks.

The paper was presented at this event. Haven't yet watched it though.

https://www.youtube.com/live/ats1HPTaV64?si=WVUDcubVINaE8K_P

Hat tip to Krzysztof Iwanek for sharing this.

Interesting discussion.

Indonesia bans social media for teens

Bhuvan — Sat, 07 Mar 2026 00:00:00 GMT

Indonesia bans social media for kids under the age of 16.

Indonesia will ban social media for children under 16, Communication and Digital Affairs Minister Meutya Hafid said Friday.

Hafid in a statement to media said that she just signed a government regulation that will mean children under the age of 16 can no longer have accounts on high-risk digital platforms, including YouTube, TikTok, Facebook, Instagram, Threads, X, Bigo Live and Roblox.

The implementation will start gradually from March 28, until all platforms fulfill their compliance obligations.

I asked ChatGPT for a summary of other countries and states that have or are thinking of imposing similar bans and restrictions:

Summary of social-media bans or major restrictions on children, by country and by state/province

This focuses on government actions aimed at banning or sharply restricting minors’ access to social media. I’ve separated places where rules are already in force or rolling out from places where they are proposed, passed but not fully in force, or blocked in court.

1) In force or rolling out

Australia

Status: In force
Rule: Platforms covered by the law must take reasonable steps to prevent Australians under 16 from creating or keeping accounts.
Start date: 10 December 2025
Notes: This is the clearest example of a national ban already in effect.
Sources:
- eSafety Commissioner — Social media age restrictions
- Australian government — Social media minimum age

Indonesia

Status: Rolling out / implementation announced
Rule: Indonesia will restrict access for children under 16 on “high-risk” platforms.
Start date: 28 March 2026
Notes: Reported platforms include TikTok, Facebook, Instagram, X, YouTube, Threads, Bigo Live, and Roblox.
Sources:
- AP — Indonesia will ban social media for children under 16
- Reuters — Indonesia to restrict social media access for children under 16

2) National governments moving toward bans or hard restrictions

France

Status: Lower house / National Assembly backed the measure
Rule: Ban children under 15 from social media.
Notes: Strong movement, but not yet best described as a fully operative nationwide ban.
Sources:
- Reuters — France’s lower house backs social media ban for those under 15
- Reuters — France’s National Assembly approves banning under-15s from social media

Spain

Status: Proposed
Rule: Ban access to social media for minors under 16.
Notes: Announced by Prime Minister Pedro Sánchez; implementation still depends on the legislative path.
Sources:
- Reuters — Spain and Greece weigh teen social media bans
- Reuters background roundup

Greece

Status: Close to announcement / under discussion
Rule: A reported ban for children under 15.
Notes: Reuters described Greece as close to announcing such a move, but not yet as having fully enacted it.
Sources:
- Reuters — Spain and Greece weigh teen social media bans
- Reuters roundup

Malaysia

Status: Planned
Rule: Government says it plans to ban social media for those under 16.
Target timing: 2026
Sources:
- Reuters — Malaysia says it plans to ban social media for under-16s from 2026

Norway

Status: Proposed / in development
Rule: Raise the relevant age threshold from 13 to 15 and work toward a legal minimum age of 15 for social media use.
Notes: This is moving in the direction of a ban, but is not yet a fully operative ban.
Sources:
- Reuters — What countries do to regulate children’s social media access

Slovenia

Status: Drafting legislation
Rule: Draft law to prohibit children under 15 from accessing social media.
Sources:
- Reuters roundup

Poland

Status: Preparing legislation
Rule: New legislation being prepared to ban social media for children under 15.
Sources:
- Reuters roundup

United Kingdom

Status: Consultation stage
Rule: The government consultation includes the possibility of banning children from social media.
Notes: This is not an enacted ban yet.
Sources:
- UK Parliament Commons Library — Children’s social media use and screen time

3) State/province-level moves within countries

Karnataka, India

Status: Announced
Rule: Ban social media use for children under 16.
Notes: Reuters describes Karnataka as the first Indian state to take this step. No implementation date was specified in the reporting I found.
Sources:
- Reuters — Karnataka bans social media for children under 16

4) U.S. states: major attempts, but many are tied up in court

Florida

Status: Passed, but enforcement has been blocked in federal court
Rule: Ban children under 14 from social media; require parental consent for 14- and 15-year-olds.
Notes: One of the strongest U.S. state laws, but its main provisions were blocked pending litigation.
Sources:
- Reuters — DeSantis signs Florida law restricting social media for people under 16
- Reuters — Judge blocks Florida law banning social media accounts for children

Utah

Status: Passed, but blocked in court
Rule: Restrictions on minors’ social media use, including age verification and tighter controls on minor accounts.
Notes: Utah is one of the earliest and most important U.S. experiments, but enforcement was blocked.
Sources:
- Reuters — Utah law restricting youth social media use blocked by judge

Ohio

Status: Blocked, later permanently struck down
Rule: Required parental consent for children under 16 to use social media.
Notes: This is an example of a state trying a softer restriction than a full ban, but the law did not survive judicial review.
Sources:
- Reuters — U.S. judge blocks Ohio law restricting children’s use of social media
- AP — Judge strikes down an Ohio law limiting kids’ use of social media as unconstitutional

Virginia

Status: Passed, then blocked in court
Rule: Limit children under 16 to one hour a day on social media, alongside age verification requirements.
Notes: This is not a full ban, but it is one of the sharpest usage restrictions tried by a U.S. state.
Sources:
- Reuters — Judge blocks Virginia law restricting social media for children
- Reuters — Virginia appeals injunction against social media time limit for children

Arkansas

Status: Blocked / struck down
Rule: Social media age-verification law aimed at stopping minors from opening accounts without parental consent.
Notes: Arkansas was an early mover, but its law was blocked and later declared unconstitutional.
Sources:
- Arkansas Advocate — Arkansas social media age verification law struck down by federal court
- Arkansas Advocate — Federal judge declares Arkansas social media age verification law unconstitutional

Bottom line

Places with the clearest real-world bans/restrictions already operating or about to operate

Australia — under-16 restrictions are in force
Indonesia — under-16 restrictions begin rollout on 28 March 2026

Places strongly moving in that direction nationally

France
Spain
Greece
Malaysia
Norway
Slovenia
Poland
United Kingdom (consultation stage only)

State/province-level examples

Karnataka, India — announced under-16 ban
Florida, Utah, Ohio, Virginia, Arkansas (U.S.) — major laws or restrictions attempted, but several have been blocked or struck down in court

One-sentence takeaway

The global pattern right now is: Australia created the most concrete template, Indonesia is the newest national mover, Europe is filling up with proposals, Karnataka is the standout subnational example in India, and the U.S. has become the main battlefield where many state-level restrictions run into constitutional challenges. ```0

On boredom

Bhuvan — Fri, 06 Mar 2026 00:00:00 GMT

Daniel Smith on boredom:

Brodsky advised his listeners to give up trying to evade the feeling, and start respecting it. Boredom exists “to teach you the most valuable lesson in your life,” he said, “the lesson of your utter insignificance.” Boredom puts us in our tiny, fragile, finite place—and thank goodness for that, for “the more finite a thing is, the more it is charged with life,” with love, pain, excitement, and fear.

If you try to distract yourself from boredom, if you run from it, all will be lost. Brodsky quoted an imperishable line from Robert Frost: “The best way out is always through.” A note written by the novelist David Foster Wallace makes a similar point: “Bliss—a second-by-second joy and gratitude at the gift of being alive, conscious—lies on the other side of crushing, crushing boredom.”

China has different attitude towards AI

Bhuvan — Thu, 05 Mar 2026 00:00:00 GMT

Chinese people aren't worried about AI:

Driverless taxis roam more than a dozen Chinese cities, and service robots putter through hotels and restaurants. Chinese tech companies have rolled out medical chatbots to help users avoid long lines at hospitals. They’ve embedded A.I. assistants throughout popular apps, so users can ask A.I. in their mapping apps to help them decide where to eat, or in their shopping apps to help them choose between pairs of shoes.

That’s because Chinese tech companies have focused intensely on real-world applications for A.I. By contrast, many leading American tech companies have been focused on more abstract goals, like developing the most cutting-edge model, or achieving artificial general intelligence.

In addition, most of China’s leading A.I. models are free to use, unlike in the United States, where users have to pay for chatbots like ChatGPT to access all their features. (In fact, Chinese companies have been giving away money and luxury cars to entice people to download their apps.)

To see a world in a grain of sand

Bhuvan — Wed, 04 Mar 2026 00:00:00 GMT

Those first 4 lines!

To see a World in a Grain of Sand And a Heaven in a Wild Flower Hold Infinity in the palm of your hand And Eternity in an hour A Robin Red breast in a Cage Puts all Heaven in a Rage A Dove house filld with Doves and Pigeons Shudders Hell thro all its regions A Dog starvd at his Masters Gate Predicts the ruin of the State A Horse misused upon the Road Calls to Heaven for Human blood Each outcry of the hunted Hare A fibre from the Brain does tear A Skylark wounded in the wing A Cherubim does cease to sing The Game Cock clipd and armd for fight Does the Rising sun affright Every Wolfs and Lions howl Raises from Hell a Human Soul

Read the full poem here.

Large language models (LLMs) are index funds

Bhuvan — Tue, 03 Mar 2026 00:00:00 GMT

Ooh, I love this analogy of LLMs being the equivalent of market beta by Karthik Shashidhar:

In a sense, LLMs are like an index with a lag. Assuming that these models are constantly trained, the “intelligence” of LLMs basically represents the average internet of the internet from a (slightly) prior point in time. And this lag is super useful in not making it a zero sum game.

As long as people use LLMs cleverly, the average quality of “intelligence” or “writing quality” can go up over time. Basically, if you consider yourself to be a bad writer, or doing something you don’t know, just use an LLM. If you know what you are doing, add your own intelligence.

In other words, LLMs are Beta. Or index investing. For any given task they will give you the average output. If you decide to vibe code an app or algorithm, what you get is the average.

I've been a long time reader of his writing and enjoy reading his musings. Subscribe to both his newsletters.

This piece reminds me of another piece by Venkatesh Rao in which he used the same analogy of LLMs as index funds:

Foundation models like GPT and Claude now serve as the index funds of language. Trained on enormous corpora of human text, they do not try to innovate. Instead, they track the center of linguistic gravity: fluent, plausible, average-case language. They provide efficient, scalable access to verbal coherence, just as index funds offer broad exposure to market returns. For most users, most of the time, this is enough. LLMs automate fluency the way passive investing automates exposure. They flatten out risk and elevate reliability.

But they also suppress surprise. Like index funds, LLMs are excellent at covering known territory but incapable of charting new ground. The result is a linguistic landscape dominated by synthetic norms: smooth, predictable, uncontroversial. Writing with an LLM is increasingly like buying the market—safe, standardized, and inherently unoriginal.

Now, as a person who works in a stock brokerage and has been a staunch advocate of index funds for over a decade, it's only natural I'd love this analogy.

By the way, I keep track of Karthik's writing on this simple site I vibe-coded to keep track of writing from writers I follow: smallweb.blog

Library in the wild

Bhuvan — Tue, 03 Mar 2026 00:00:00 GMT

What an inspiring man and what an inspiring story:

“I was facing this issue for a long time. The elephants were getting more aggressive. I lived in fear of them attacking the library. Finally, I decided that if I shifted it, I would also move my house. Protecting the knowledge the library gave was important.” He did not want the learning it stood for to be disrupted.

So he shifted the library a few kilometres, though still within the forested region of Edamalakudi. “The very week I left,” he says, tuskers “demolished the whole old structure in minutes.” But, he adds proudly, “I’d moved all the books out of there.” Being a forest dweller, he understood that something was going on with the great beasts. And bears no bitterness towards them. Moving, though, was a wise decision as they ravaged the entire structure. And Chinnathambi had no wish to literally discover the elephant in the room.

Misunderstanding China's ambitions

Bhuvan — Tue, 03 Mar 2026 00:00:00 GMT

💯

China is quite aggressive and territorially self-aggrandising in its immediate backyard, but restrained beyond. Even occasional border skirmishes with India are basically fought with sticks and stones. Outside of the military build up triggered by the US during MAGA 1.0, China is not interested in projecting power globally. And why would they? The legitimacy of its political class is not bound up in maintaining global hegemony.

You have to inhabit a different plane of existence if you think those aspects of China’s foreign policy warrant a comparison to the full-spectrum vandalism of the US and its satellites which has left hundreds of thousands of corpses and ruined countries in its wake

Also see: What Does China Want?

Our findings indicate that China is a status quo power concerned with regime stability and is more inwardly focused than externally oriented. China's aims are unambiguous, enduring, and limited: It cares about its borders, sovereignty, and foreign economic relations. China's main concerns are almost all regional and related to parts of China that the rest of the region has agreed are Chinese—Hong Kong, Taiwan, Tibet, and Xinjiang. Our argument has three main implications. First, China does not pose the type of military threat that the conventional wisdom claims it does. Thus, a hostile U.S. military posture in the Pacific is unwise and may unnecessarily create tensions. Second, the two countries could cooperate on several overlooked issue areas. Third, the conventional view of China plays down the economic and diplomatic arenas that a war-fighting approach is unsuited to address.

The brutal economics of music streaming

Bhuvan — Mon, 02 Mar 2026 00:00:00 GMT

I don't know enough about the music industry to know if this piece is true, but if it's true, then it's depressing.

In a normal tech business (like SaaS or Netflix), as you gain more subscribers, your profit margins increase exponentially because your fixed costs stay relatively stable. Once Netflix pays $20 million to produce an original movie, that cost is fixed. Whether 1 million or 100 million people watch it, the cost doesn’t change. The margin expands.

Streaming music operates in reverse. Because DSPs pay out roughly 70% of every dollar earned back to rightsholders (labels and publishers), their costs scale linearly with their user base. Every time a song is streamed, a fraction of a cent leaves the building.

Iovine put it bluntly: “The streaming services have a bad situation, there’s no margins, they’re not making any money.”

This model only works for Apple, Amazon, and Google, because they don’t need their music platforms to be wildly profitable. Amazon uses music as a loss-leader to keep you paying for Prime. Apple uses it to sell $1,000 iPhones.

Too distracted, too bored, and too stupid to care?

Bhuvan — Mon, 02 Mar 2026 00:00:00 GMT

Too distracted, too bored and too stupid to care? Brilliant piece by Lily Lynch:

I posted an observation on X yesterday that generated a lot of talk and hate. I wrote that I’d seen a lot of people complaining that there was no propaganda campaign in the lead-up to this war, no effort to convince the American public that we were at imminent risk. There was a rather haphazard, tacked-on attempt by Steve Witkoff to advance the idea that Iran was a “week away” from being able to build a nuclear bomb, but this messaging came late and contradicted the administration’s prior reports about an obliterated nuclear program. Indeed, the Trump administration’s foreign policy is warned over boomer neoconservativism, but unlike the neocons before him, he hadn’t tried to give the war on Iran a clearly defined objective or coherent propaganda narrative. On X, I suggested that this was because he no longer has to: there have been no mass anti-war protests like the kind that accompanied the runup to the invasion of Iraq. Several people objected to this, claiming that protests in 2003 only started after the war began, but this, as anyone who participated in those protests remembers, is false: there were massive protests against the war before even a single missile was launched. I suggested that the reason the government can count on our relative inaction is that the American people are largely pacified by technology, utterly stupefied and numb from online shopping and personalized algorithms mainlining a steady stream of mind-melting short-form video content to each individual. Of course, there are still pockets of virulent opposition on college campuses and in the streets which I don’t wish to diminish, but in wars past, protest was truly a mass phenomenon. And I do think that we are more atomized and more pacified today. Back in 2003, we hadn’t yet completely retreated into our screens.

Use the damn tools and then judge

Bhuvan — Sun, 01 Mar 2026 00:00:00 GMT

It’s kind of fun to watch many AI skeptics move from pure, unbridled skepticism about the usefulness of AI tools like Claude Code to becoming, if not outright boosters, at least genuine admirers of them.

And this is just one example. Across many such cases, the pattern I keep seeing is that people often have a terrible model for judging these tools.

First, a lot of people simply regurgitate other people’s opinions: that these tools are bad, that they hallucinate, that they are “stochastic parrots,” or whatever other cliché they picked up from a tweet, a video, or somewhere else.

Second, many people don’t actually use the latest cutting-edge tools and models. They either stick to the web interfaces, which are good but still limited in what they can do, or they use older models and then jump to sweeping conclusions.

But until people actually use the latest tools, like Claude Code, OpenAI Codex, or Cursor, they won’t really understand how far these systems have come. They won’t see just how remarkable these tools have become, how good they are at helping people solve not just mundane problems but genuinely complicated ones, and how much they can enhance human capability.

Here's Chris Blattman:

4w ago I was a Claude Code skeptic. I'm not a coder. None of the use cases were relevant. I managed teams & projects, drowning in email & overdue reminders. So I tried creating tools that would help me and... holy crap.

Now I'm sharing the tools I built: claudeblattman.com

Terrence Tao on AI

Bhuvan — Sat, 28 Feb 2026 00:00:00 GMT

Terrence Tao on what AI can and cannot do:

Tao: Today there are a lot of very tedious types of mathematics that we don’t like doing, so we look for clever ways to get around them. But AIs will just happily blast through those tedious computations. When we integrate AI with human workflows, we can just glide over these obstacles.

I also think mathematicians will start doing math at larger scales. Think about the difference between case studies and population surveys in sciences. If you were to study a disease in the 18th century, if it was a rare disease, you might study one patient who has this disease and record all their symptoms and take meticulous notes. But in the 21st century, you can do a clinical trial and you can administer a drug to 1,000 people and do statistics and get much more precise information about the efficiency of your drug.

Mathematics is still very much at the case-study level. A paper will take one or two problems and study them to death in a very handcrafted, intensive way. That’s our style. But what AI tools enable is population studies.

AI-powered power review

Bhuvan — Fri, 27 Feb 2026 00:00:00 GMT

John Cochrane gushing about Refine, the AI-powered research tool:

This is the first time I’ve seen AI at work in something I do daily, and it really is revolutionary.

Refereeing and evaluating papers is one of the more unpleasant and time consuming tasks in our profession. I’ve read a lot of referee reports in my 40 years as an economist, and this is top 5% for sure. Most referee reports do not identify the major point of the paper, and do not assess if the paper backs up that point. They do not notice glaring gaps of logic, basic theorems violated, econometrics advice 101 ignored. Editors are lucky if one out of three reports is vaguely useful. Clearly, this task is going to be radically impacted by AI. If I were an editor, I’d feed every paper to refine on receipt. Or, I would require the author to spend the $50 and send in the last refine report! I will surely get refine’s opinion before any referee report I write in the future.

Will all the referees be out of jobs? No! You still have to read and evaluate what refine offers. But the speed, accuracy, and quality of reports will jump. And economists will save a lot of time.

Cognitive war

Bhuvan — Fri, 27 Feb 2026 00:00:00 GMT

The backlash against social media is gaining ground. I wouldn't be surprised if there are widespread bans across the world in some shape or form in the next couple of years.

An interesting take on the issue:

China knows exactly what this does. Their domestic version of TikTok caps kids under 14 at 40 minutes a day, locks access between 6am and 10pm, and swaps the entire feed to educational content. Science, history, museums.

The version they export to everyone else? Unlimited, unrestricted, pure dopamine on demand.

When kids in the US and China were asked what they wanted to be when they grew up, the number one answer in America was influencer. In China it was astronaut.

Macron calls this a cognitive war. Export what dulls young minds and keep what makes them intelligent for your own population.

This is the most effective weapon ever deployed against a generation’s ability to think.

Jeremy Siegel on an AI Apocalypse

Bhuvan — Fri, 27 Feb 2026 00:00:00 GMT

Jeremy Siegel on the Citrini piece and the looming AI apocalypse:

Before the industrial revolution, humankind had to work 12 hours a day, seven days a week to produce the goods for bare-bones survival: food and basic shelter. Today, in the modern world, survival can be achieved by working less than a day. In 1800, 80% of American workers toiled in the farming industry. Today that is less than 2% and the US exports food to the rest of the world.

Pessimists maintain that the speed of the AI Revolution makes these historical comparisons irrelevant (although I remember predictions that Amazon and the internet would eliminate all retail by the middle of the 2020s). But that assertion fails to take account of the great flexibility of workers to adapt to sudden changes. Shifts in production during wars and even the recent pandemic showed how economies can adapt to changes in circumstances far faster than those induced by AI. Workers with highly developed skills will find great demand from the new industries that will be enabled by artificial intelligence.

AI will not bring an economic apocalypse. AI will lead us to the greatest increase in the standard of living since the dawn of the Industrial Revolution. Although some professions will clearly be impacted, the dramatic increase in consumer income will more than offset the loss of jobs caused by the new technology. The AI future is not one to be feared, but one to be embraced.

The physical limits of AI

Bhuvan — Thu, 26 Feb 2026 00:00:00 GMT

From Citadel's response to the Citrini piece:

Furthermore, it is well acknowledged that training and inference requires significant semiconductor capacity, data centers, and energy. Displacing white collar work would require orders of magnitude more compute intensity than the current level utilization. If automation expands rapidly, demand for compute definitionally rises, pushing up its marginal cost. If the marginal cost of compute rises above the marginal cost of human labor for certain tasks, substitution will not occur, creating a natural economic boundary. This dynamic contrasts sharply with narratives assuming frictionless replication of intelligence. Even if algorithms improve recursively, economic deployment remains bounded by physical capital, energy availability, regulatory approvals, and organizational change. Recursive capability does not imply recursive adoption. The Citrini piece.

The stupidty of luxury goods

Bhuvan — Thu, 26 Feb 2026 00:00:00 GMT

That humans—supposedly the smartest creatures in the known universe—splurge thousands of dollars on pointless baubles all to get the validation of strangers never ceases to amaze me:

Even as luxury goods have become unattainable for those French and Italians meant to desire them, they have become near-ubiquitous for the foreign moneyed masses. Luxury houses sell the idea of scarcity, with hordes of publicists explaining that the years-long wait for a Birkin handbag is due to the lack of sufficient artisans to craft these pinnacles of refinement. This is a fairy tale stitched in fine silk. The luxury-goods industry has roughly tripled in size since 2000; its €358bn in annual sales—half a Walmart or Amazon, give or take—betrays how thoroughly mainstream supposed exclusivity has become. Fifty years ago, Louis Vuitton had but two outlets, both in France. These days it has two stores in Ningbo, China’s 34th-biggest city. Exclusive, moi?

Perhaps one should laud Europeans, who saw a group of mugs willing to wait years for the right to buy a €50,000 Swiss watch and obligingly took their money. Once upon a time, the church in Europe sold indulgences to shorten the buyers’ stays in purgatory. Now luxury’s high priests sell trinkets meant to shorten their clients’ stay in social obscurity. America exports F-35s. South Korea exports K-Pop. Europe has found a way to export self-worth.

Why Is Fertility So Low in High-Income Countries?

Bhuvan — Thu, 26 Feb 2026 00:00:00 GMT

Saving this paper to read later:

We tentatively conclude that the decline in fertility across recent cohorts in high-income countries is likely the result of evolving norms, expanding choices, and the role of broad social and economic influences. This multi-faceted cohort-based explanation contrasts with narrower period-specific explanations focused on the role of prices, income, and the opportunity cost of women’s time. Those standard types of explanations are likely important at a micro level, but to understand the widescale decline in fertility and rise in childlessness across recent cohorts requires a wider lens.

Beyond this tentative conclusion to the posed title question, we offer the following additional suggestions for future research among economists interested in this area of study - p. 39 First, future work in economics into the decline in fertility would benefit from more fully integrating demographic insights. As noted above, demographers have often focused on the mechanics of fertility trends and have long emphasized the importance of understanding lifecycle patterns in fertility at the cohort level, including distinctions between timing, total number of children ever born, and childlessness. Economists often emphasize the causal behavioral determinants of fertility outcomes. Both perspectives are essential for a comprehensive understanding of fertility patterns and the integration of the two is likely to be the most productive path forward for advancing understanding.

There are no solutions. There are only trade-offs

Bhuvan — Wed, 25 Feb 2026 00:00:00 GMT

I keep thinking about this quote:

“There are no solutions. There are only trade-offs.” ― Thomas Sowell

Parody is now reality

Bhuvan — Tue, 24 Feb 2026 00:00:00 GMT

This is not from The Onion but from the Financial Times:

Officials working with Donald Trump’s “Board of Peace” are exploring setting up a stablecoin for Gaza as part of efforts to reshape the devastated Palestinian enclave’s economy, according to five people familiar with the discussions.

The talks around introducing a stablecoin — a type of cryptocurrency whose value is pegged to a mainstream currency, such as the US dollar — are at a preliminary stage, and many details of how one could be introduced in Gaza remain to be determined.

AI as an excuse to fire people

Bhuvan — Mon, 23 Feb 2026 00:00:00 GMT

I hate to agree with Altman but I think he's right:

“I don’t know what the exact percentage is, but there’s some AI washing where people are blaming AI for layoffs that they would otherwise do, and then there’s some real displacement by AI of different kinds of jobs,” Altman said

Another dodgy self help guru?

Bhuvan — Sun, 22 Feb 2026 00:00:00 GMT

I know of Jay Shetty, but I've never heard his podcast or read his books. I know that he's quite popular, and there have been questions about his past. What fascinates me is why such people who loosely preach some form of "self-help" become popular despite the questionable nature of their advice, their disingenuousness, and the outright grift. I'm not saying Jay Shetty is a fraud – I don't know. But I'm fascinated by people like him.

What fascinates me is the continued popularity of self-help despite the painfully bland advice. The fact that books that tell you to fold your blanket after you wake up and wash your ass after taking a shit continue to be bestsellers is funny and tragic at the same time.

I get that these books, talks, and podcasts make it seem like they have answers and there's money to be made in exploiting people's hopes, fears, dreams, insecurities, and anxieties. Hell, I work in finance and I see everyday how base emotions drive people mad, but I'm still fascinated nonetheless. Being a self-help guru is probably the only AI-proof job?His recent interview in The New York Times and an older piece in The Guardian.

Nate Silver on the political impact of artificial intelligence (AI)

Bhuvan — Sun, 22 Feb 2026 00:00:00 GMT

Agree with everything Nate Silver is saying in this post about the cluelessness and idiocy of both Silicon Valley and politicians about the potential fallout of AI if it indeed is a big deal:

Disruption to the “creative classes” could produce an outsized political impact. I’m not exactly sure where people in the creative classes — say, writers or editors or artists [6] or, to broaden the net, industries like consulting or advertising — rank in terms of the medium-term threat from AI-related job displacement. (Journalists per se have long lived with an anvil over their heads in a perpetually struggling industry.) These are the people I tend to hang out with. In our darker moods, we sometimes have conversations about who will or won’t have a job in five years. I suspect they’re at above-average risk, though — less threatened than, say, mediocre programmers but more than, say, someone with irreplaceable physical gifts like Victor Wembanyama. However cynical one is about the failings of the “expert” class, these are people who tend to shape public opinion and devote a lot of time and energy to politics. If a consensus develops among this cohort that their livelihoods are threatened, or that their children’s livelihoods are, I expect there will be enough political blowback that anti-elite pushback won’t be enough to overcome it.

Chesterton fences everywhere

Bhuvan — Sat, 21 Feb 2026 00:00:00 GMT

What a beautiful article. It captures two ideas that I often think about.

First idea: The sheer extent of our ignorance. We go through life thinking we know a lot, but in reality, what we know isn't even a drop in the cosmic ocean. Yet thinking we know more than we actually do serves an evolutionary purpose—it prevents us from being crippled by anxiety, doubt, and fear. It gives us the confidence to act and move forward. The problem is that this same confidence, without the humility to recognize what we don't know, can often lead to disasters.

Second idea: We don't truly appreciate things until they're gone—and a related version: extraordinary things become commonplace very quickly for humans. This adaptation helps us because we can't have our minds constantly blown by wonder. But there's a trade-off: when things lose their magic through familiarity, we take them for granted. We stop maintaining them, stop tending to them. We neglect the responsibility required to keep them from deteriorating and falling apart—to fight against entropy.

Readers might recall GK Chesterton's famous fence, a warning against the destruction of existing systems without first understanding their purpose:

There exists in such a case a certain institution or law; let us say, for the sake of simplicity, a fence or gate erected across a road. The more modern type of reformer goes gaily up to it and says, “I don’t see the use of this; let us clear it away.” To which the more intelligent type of reformer will do well to answer: “If you don’t see the use of it, I certainly won’t let you clear it away. Go away and think. Then, when you can come back and tell me that you do see the use of it, I may allow you to destroy it.”

Chesterton's cautionary tale speaks to this good-fitness problem.

When a received solution works well, we don't notice its positive function. The proverbial fence was built by someone long ago to serve a purpose, one it served so silently that the reasons for its being have been lost. The fence sits there doing its job, downstream benefits taken for granted.

Chesterton fences everywhere:

The U.S. entrepreneur Steve Blank gave a great example of Chesterton Fence’s in action. Imagine that a successful startup is growing, and so they hire a Chief Financial Officer to get their ducks in a row. The new CFO, keen to show intent, finds some cuts to make and costs to save. And so they get rid of the free soda and snacks in the office. It’s a job well done. That’s $10,000 saved. But suddenly, the company feels very different. The employees who made the startup successful start to mutter and gripe. “Things are too corporate now,” they say, and “it’s not like it was.” So, an exodus happens. They leave for greener pastures. They leave the cost-cutting ship. They miss their free soda.

How India gives

Bhuvan — Sat, 21 Feb 2026 00:00:00 GMT

$6 billion is a lot for a poor country like India:

The How India Gives 2025 report, produced by the Centre for Social Impact and Philanthropy (CSIP) at Ashoka University, challenges the conventional narrative that organised, institutional money dominates the country's giving landscape. Instead, it points to a quieter colossus: households.

According to the report, India's total household giving is estimated at 540bn rupees ($6bn) annually, including cash, in-kind contributions and volunteering.

About 68% of respondents report giving in some form. Of this, 48% is in kind - such as food, clothing or other household goods - followed by cash donations (44%) and volunteering (30%) with non-profits, religious institutions or community groups.

Much of the food given goes to communal free kitchens. Volunteering most commonly takes the form of service at religious institutions, including activities such as disaster relief organised by them.

"India is a very generous country. Our findings suggest that ordinary households play a much larger role than is commonly acknowledged. Generosity appears widespread and culturally embedded," Jinny Uppal, head of Centre for Social Impact and Philanthropy at Ashoka University, told the BBC.

Trump's stupid trade war takes a stupid new turn

Bhuvan — Sat, 21 Feb 2026 00:00:00 GMT

The US Supreme Court says that Trump's tarrifs are illegal and Trump finds a new way impose the same tariffs.

The court’s 6-3 decision has significant implications for the U.S. economy, consumers and the president’s trade policy. The Trump administration had said that a loss at the Supreme Court could force the government to unwind trade deals with other countries and potentially pay hefty refunds to importers.

Mr. Trump is the first president to claim that a 1970s emergency statute, which does not mention the word “tariffs,” allowed him to unilaterally impose the duties without congressional approval.

Writing for the majority, Chief Justice John G. Roberts Jr. said that statute does not authorize the president to impose tariffs.

“The president asserts the extraordinary power to unilaterally impose tariffs of unlimited amount, duration, and scope. In light of the breadth, history, and constitutional context of that asserted authority, he must identify clear congressional authorization to exercise it,” the chief justice wrote.

In response Trump calls the judges...

The justices were “fools and lapdogs” as well as “unpatriotic and disloyal”, Trump said at a hastily arranged press conference at the White House. The president alleged that they were “swayed by foreign interests” and said other countries were “dancing in the streets”.

“I’m ashamed of certain members of the court, absolutely ashamed for not having the courage to do what’s right for our country,” he said.

Oh and hes doubling down on his stupid trade war:

President Donald Trump on Friday evening said he signed an executive order imposing a new 10% “global tariff,” hours after the Supreme Court struck down his sweeping “reciprocal” import duties in a major rebuke of his trade agenda.

The “Section 122” tariffs will take effect “almost immediately,” Trump said in a Truth Social post.

Trump has other options to continue the dumbest trade war in the history. They are somewhat bad but he can continue his stupidity unchecked. Doug Irwin in The Economist:

Although the broad discretionary and arbitrary scope of the IEEPA approach may be gone, many other cudgels remain. The president has ample authority under other statutes to rebuild the tariff wall, albeit with somewhat greater difficulty and many more holes.

Some of these statutes allow for across-the-board tariffs on all goods from all countries; Section 122 of the Trade Act of 1974, for instance, allows the president to impose tariffs of up to 15% for up to 150 days for balance-of-payments reasons. These are supposed to be non-discriminatory. Some are more narrowly targeted at particular countries engaged in unfair trade practices. Section 301 was used to impose tariffs against China in Mr Trump’s first term and could be brandished against other newly declared unfair traders. Some are aimed at stopping imports of specific goods from any country, such as Section 232 of the Trade Expansion Act of 1962 on national-security grounds. These have already been deployed in the case of steel, aluminium, semiconductors and other products—and the Trump administration is looking to impose even more.

Geopolitical survival guide

Bhuvan — Fri, 20 Feb 2026 00:00:00 GMT

Geopolitical volatility doesn't matter if you are an investor:

The knee-jerk reaction of investors to geopolitical crises is to extrapolate the most recent events into the future and expect an escalation of a new conflict. This is when geopolitical experts are on TV and in the press with their predictions of World War III or a 1970s style oil shock and stagflation, etc.

Ignore those warnings and doom and gloom forecasts as in most cases they are based on an assumption that the crisis rapidly escalates and gets out of control. However, that rarely happens. Over the last 150 years we have seen only two instances where wars got out of control. They are called World War I and World War II.

But we have seen hundreds of instances when a war broke out that could conceivably trigger World War III but hasn’t. Think of the Korean War and the Vietnam War, the Cuban Missile Crisis, the many wars in the Middle East or the constant tensions between a nuclear armed North Korea and its neighbours. Think of the many civil uprising during the Arab Spring of 2011. Yet, things rarely get out of control because, well, people like to live in peaceful times and will do their best to avoid going to war. It requires staggering miscalculation on both sides to escalate a war.

A promise is a direction taken

Bhuvan — Thu, 19 Feb 2026 00:00:00 GMT

A promise is a direction taken, a self-limitation of choice. As Odo pointed out, if no direction is taken, if one goes nowhere, no change will occur. One's freedom to choose and to change will be unused, exactly as if one were in jail, a jail of one's own building, a maze in which no one way is better than any other.

-

You don't understand what time is,' he said. 'You say the past is gone. the future is not real, there is no change, no hope. You think Anarres is a future that cannot be reached, as your past cannot be changed. So there is nothing but the present, this Urras, the rich, real, stable present, the moment now. And you think that is something which can be possessed! You envy it a little. You think it's something you would like to have. But it is not real, you know. It is not stable, not solid—nothing is. Things change, change. You cannot have anything. And least of all can you have the present, unless you accept with it the past and the future. Not only the past but also the future, not only the future but also the past! Because they are real: only their reality makes the present real. You will not achieve or even understand Urras unless you accept the reality, the enduring reality, of Anarres.

Ursula K. Le Guin, The Dispossessed: An Ambiguous Utopia

Came across this quote in this post.

An AI reading companion

Bhuvan — Wed, 18 Feb 2026 00:00:00 GMT

Having an AI assistant while reading has been really helpful. It's like having a person who knows a lot about a lot of things at your beck and call. Considering the fact that LLMs like ChatGPT and Claude are pretty much trained on all the written knowledge of humanity available on the internet, they make for excellent teachers. So I built this simple eBook reader with an AI integration.

You can upload an eBook, and if you are having trouble understanding anything, you can invoke an LLM of your choice and then ask questions. All you have to do is set an API key using Openrouter or your LLM of choice.

Download it here

What’s wild is that I pointed out a bunch of things to OpenAI’s Codex, and it just did them—suggested a bunch of other interesting and helpful features, figured out a bunch of bugs in the editor, and just fixed them on its own.

And I have absolutely no clue how this entire thing works because I’m not a coder. But even thinking—I mean, I’ve said this before on the same blog—the fact that I could even think about doing this, let alone actually do this, itself is very, very wild.

And so I was just reading some of the book, and I was actually using this to keep my own cooking, and I kind of started figuring out a bunch of things. And I kind of started noticing these small little improvements in the reading experience and the chat, and Codex is like, “No worries, we’ll just do it.” And it just kept on doing it.

I mean, this is not a serious piece of software. This is something I just built to solve my own issue and to improve my reading experience—improve my own… to help me read more.

And it’s kind of crazy that this is even possible. Wild, wild times.

AI-assisted coding is the new entertainment

Bhuvan — Tue, 17 Feb 2026 00:00:00 GMT

Funny but probably true? A sign of things to come?

The anxiety is rational, which is why it sticks. Every week some new benchmark drops that makes last month's workflow feel prehistoric. Codex ships overnight processing. Opus gets faster. Context windows double. None of it reduces the pressure. It multiplies it. You can do more now. And someone already is. The window to be first at anything feels like it's shrinking by the day. Literally, by the day.

I replaced Netflix with Claude Code. I lie in bed thinking about what I can spin up before I fall asleep, what can run while I'm unconscious. Reading a novel feels indulgent now. Watching a movie without a laptop open feels wasteful. This voice in my head that says "something could be running right now" just doesn't shut off. I'm not even building a company. I'm just addicted to building my random ideas.

Technology saturated beings

Bhuvan — Tue, 17 Feb 2026 00:00:00 GMT

It's an obvious, but underappreciated, fact that technology reshapes reality. But we don't often think about it much less care about it. If you even spend a minute thinking about just how much of the shape of our lives is sculpted by the technologies we use and rely on, your mind will be blown, and I'm not even talking about AI – just all the regular technologies we use from Google Maps, Messenger apps, smartphones, and thermostats.

This came to my mind as I was reading this thoughtful article by Alastair Benn:

The irony is that modern life has made redundant many of the routines which were once indispensable to passing language oral examinations. Buying a ticket at a railway station is now either entirely automated or conducted online. The last time I bought a ticket in person, I did so purely for the novelty of stating my destination to another human being, and at what date and time I intended to leave and return. The number of moments in daily life that require people to articulate themselves in a shared, formal register has shrunk markedly, even compared with the recent past.

The fundamental problem is complicity

Bhuvan — Tue, 17 Feb 2026 00:00:00 GMT

Brutal but true:

Hannah Arendt wrote that if the fundamental problem of postwar intellectual life after 1918 was death, then after 1945 it was evil. Today, the fundamental problem is complicity: reckoning with our acquiescence to a political system that suddenly appears criminal. These shocks all expose something rotten beneath the surface of liberal respectability: how elites participate in atrocity while maintaining prestige; how institutions supposed to safeguard justice in fact protect the powerful and exploit the vulnerable; how mass consent is secured through careful distribution of access and silence. The names in those Epstein files – political leaders, corporate titans, academics and cultural impresarios – are disturbing because their continued authority and uninterrupted careers make undeniable what we have known but perhaps refused to fully concede: that the moral vocabulary of the US-led liberal international order became a cover for domination and kleptocracy.

The Epstein outrage and Trump’s demolition of American prestige have alike led the collapse of enforced silence. Mark Carney’s admission at Davos that the US-led order had passed was applauded by the very Atlanticist elites who championed it for decades. Trump’s obscenity has made the system impossible to defend with the usual pieties. That these same commentators were unmoved by Biden’s sponsorship of genocide shows what actually troubles them: not the order’s concealed violence, but the loss of its dignified facade. It is now a scramble to salvage reputations amid general collapse.

Iain McGilchrist on attention

Bhuvan — Mon, 16 Feb 2026 00:00:00 GMT

Beautiful meditation on attention:

McGilchrist considers the way our attention constructs our reality and becomes the beating heart with which we love the world:

The whole illuminates the parts as much as the parts can illuminate the whole… The world we experience — which is the only one we can know — is affected by the kind of attention we pay to it.

Defining attention as “the manner in which our consciousness is disposed towards whatever else exists,” he writes:

The choice we make of how we dispose our consciousness is the ultimate creative act: it renders the world what it is. It is, therefore, a moral act: it has consequences.

Go to the limits of your longing

Bhuvan — Sun, 15 Feb 2026 00:00:00 GMT

I woke up and began my doomscrolling session on Substack and came across a note by the awesome Cate Hall with a screenshot of Rilke's 'Go to the Limits of Your Longing' poem. The translation is under copyright but out of curiosity I asked Gemini and Claude to find the original German version and then translate it:

God speaks to each of us before our making, then walks in silence as the night is breaking. But the words spoken before we each begin— these words, like silver, are these:

You, sent out beyond your own senses, go to the edge of your longing; clothe me.

Grow like a fire behind all things, so that their shadows, stretched wide, may always cover me entirely.

Let everything happen to you: beauty and terror. One must only keep going: no feeling is the furthest. Do not let yourself be severed from me.

Nearby is the land that they call life. You will recognize it by its gravity.

Give me your hand.

I know it’s become trivial to use AI tools like ChatGPT or Gemini, but the fact that I can ask a tool to find a German poem, translate it into English, and then help me understand it—all from my phone—is still kind of crazy.

We are all unhappy wanderers

Bhuvan — Sun, 15 Feb 2026 00:00:00 GMT

This post by Bruce Sorkin hit hard because I'm most often than not a tourist in the present, with permanent residences in the past and the future:

Csikszentmihalyi talked about challenge, but I would also here invoke the Japanese concept of Beginner’s Mind or Shoshin. Derived from Zen Buddhism, and particularly popularized in the West by Shunryu Suzuki, it refers to dropping our preconceptions and approaching a subject with openness, eagerness, and a lack of bias—just as a beginner would. When your attitude is one of openness and exploration rather than merely confirming that the experience meets expectations, there is a greater chance of becoming absorbed into the experience, having your thoughts conform to the contours of the experience, and entering flow. You can approach an experience with wonder even if it is familiar.

A Mind Off the Leash is a lost mind—scared, wandering, and at risk.

A Mind on a Short Leash is a strained mind—rigid, anxious, forced and rebellious.

A Mind on a Soft Leash is a present mind—guided by Shoshin, open to the Flow, and truly at home.

An emotional wreck

Bhuvan — Sat, 14 Feb 2026 00:00:00 GMT

This interview of Daniel Radcliffe is wonderful. It reminded me of this line that's been rattling in my head: "We're all stumbling and suffering through life on how to live life.”

You’ve described Every Brilliant Thing, where your character makes a list of reasons to live, as “a very funny play about depression.” What would be on your list? Most of mine at the moment would probably be about my son — all the bizarre pronunciations he has for things. I took him to [the live monster-truck show] Monster Jam, and there’s a truck that’s a unicorn, and it shoots confetti out of its horn. The truck is called Sparkle Smash, if you’re interested. I heard him telling that story later, and he referred to the confetti as “spaghetti.” Also, his godfather gave him a little toy-car model of a Lamborghini, which he has remembered as “zucchini.” So there’s any number of just incredibly sweet things that he says.

I’d also say the music of Joanna Newsom or Tom Lehrer. I could do a list of a thousand brilliant things that were just records. And there’s one in the show that I love, which is “Waking up late with someone you love.” That is a brilliant thing in life.

What’s the best advice you’ve ever gotten? Something my dad said has always stuck with me. “When you’re meeting people, just get your hand out first.” I’ll go into a room full of a lot of people and will feel a deep need to shake everybody’s hand. As a kid, I was so aware of the stereotype around child stars. I was like, “Man, everyone thinks I’m going to be a dick everywhere I go. I just have to try and prove as quickly as possible to somebody that I’m not.” It’s a way to show that you’re not an asshole.

Social media bans are hard to enforce

Bhuvan — Sat, 14 Feb 2026 00:00:00 GMT

The Economist magazine makes the case that social media bans won't work as well as people think they do. Im fascinated by this debate and don't have enough understanding of the nuances to have a good view yet.

Moreover, the proponents of bans ignore how they would deprive children of the benefits of social media. They are a blessing to children who feel isolated: perhaps because of their location, their sexuality, or because their brains work differently from those of others. Social media can broaden young minds, giving children from all backgrounds a window onto fresh places and people. Like it or not, social sites are now one of the main ways children obtain information (as well as misinformation) about current affairs. It used to be easy for youngsters to pick up their parents’ printed newspapers. They sat through news bulletins aired before or after their favourite shows. Those days are no more.

The > Teenagers who are turfed off TikTok will not instantly begin climbing trees or poring over books. Many will slump for longer in front of games consoles and streaming services. One reason they spend so many hours online is that parents long ago stopped letting them hang around outside with friends. Having chased them indoors, adults should now think twice about placing further prohibitions on their free time.

Simone Weil is awesome

Bhuvan — Fri, 13 Feb 2026 00:00:00 GMT

I was doing some research for a post and came across the great Simone Weil. I had heard her name and that's about it. I just started listening to a few podcast epsiodes about her and I'm already becoming a fan of the remarkable life she led and the range and density of her thinking.

https://youtu.be/4RkMns4ZXhc?si=8AjAIwre97g0gmON

Things we lost in the fire

Bhuvan — Thu, 12 Feb 2026 00:00:00 GMT

A developer with 40 years of experience on how it feels like to program in the shadow of AI:

I don’t want to be dishonest about this. There’s a version of this post where I tell you that experience is more valuable than ever, that systems thinking and architectural judgment are the things AI can’t replace, that the craft endures in a different form.

And that’s true. When I’m working on something complex — juggling system-level dependencies, holding a mental model across multiple interacting specifications, making the thousand small decisions that determine whether something feels coherent or just works — I can see how I still bring something AI doesn’t. The taste. The judgment. The pattern recognition from decades of seeing things go wrong.

AI tools actually make that kind of thinking more valuable, not less. When code generation is cheap, the bottleneck shifts to the person who knows what to ask for, can spot when the output is subtly wrong, and can hold the whole picture together. Typing was never the hard part.

But I’d be lying if I said it felt the same. It doesn’t. The wonder is harder to access. The sense of discovery, of figuring something out through sheer persistence and ingenuity — that’s been compressed. Not eliminated, but compressed. And something is lost in the compression, even if something is gained.

We don't know how it all works

Bhuvan — Thu, 12 Feb 2026 00:00:00 GMT

Lorin Hochstein responds to people who worey that AI "generates" code that we understand and itll probably lead to disasters down the line:

Finally, Bucciarelli is right that systems like telephony are so inherently complex, have been built on top of so many different layers in so many different places, that no one person can ever actually understand how the whole thing works. This is the fundamental nature of complex technologies: our knowledge of these systems will always be partial, at best. Yes, AI will make this situation worse. But it’s a situation that we’ve been in for a long time.

Build something you care about and benefits others

Bhuvan — Wed, 11 Feb 2026 00:00:00 GMT

I became a huge fan of Zena Hitz ever since I heard her conversation with Henry Oliver. After listening to her, I immediately bought her book but haven't read it yet (sorry Zena), but it's on my reading list for this year. I just finished reading this lovely interview of hers and this stuck with me:

Everyone seems to want “scale” and “impact” and seems to think that means having a massive operation. But doing something right on a small scale breeds imitators—it is like a seed. Great Books institutions were always small, but they have been enormously influential. There’s nothing more “scalable” than wheat or bread, but no one owns all the wheat nor the bread-making recipes. It doesn’t take much thinking to see that this is for the best.

Technology only works for human flourishing if we choose to design it and to use it that way. We need to think about how we want to live—in common as well as individually—and choose accordingly. Likewise, money isn’t the most important thing. To build trust, you need to care about something that recognizably benefits more than just you and your friends.

The shadowy world of abandoned oil tankers

Bhuvan — Tue, 10 Feb 2026 00:00:00 GMT

Big spike in abandoned vessels. A lot of these ships must be the same ships that were used by Russia to skirt western sanctions and continue shipping oil:

Back in 2016, 20 ships were abandoned around the world, according to the ITF. In 2025 the number had ballooned to 410, with 6,223 merchant seamen falling victim. Both of those figures for last year were up by almost a third on 2024.

Geopolitical instability is said to have been a driving factor of the increase in recent years. Widespread conflicts around the world and the Covid pandemic have triggered supply chain disruption and wild variation in freight costs, meaning some operators are struggling to stay afloat.

But the ITF says the growing prevalence of so-called "shadow fleets" could be contributing to the big spike last year.

These ships, typically oil tankers such as the one Ivan is stuck on, are more often ageing vessels of obscure ownership, unseaworthy, likely uninsured, and operationally hazardous. And they typically sail under flags of convenience or FOCs - the ships are registered in countries with very limited regulatory oversight.

Pratap Bhanu Mehta on the India-US trade deal

Bhuvan — Tue, 10 Feb 2026 00:00:00 GMT

I'll resist quoting Thucydides. As usual, anothet brilliant perspective by PB Mehta:

But even if we acknowledge that the agreement has some pragmatic economic potential, it does not pass the smell test. For starters, this is not a reciprocal free-trade agreement. As Trump has made clear, America is not playing for reciprocal equality; it is playing for imperial domination. Even on its own terms, the agreement reflects this asymmetry. The new tariff regime is worse for India than what existed before Trump came to power. In a manner reminiscent of 19th-century imperial trade, the tariff structure favours the United States: India cuts tariffs to zero, while the US imposes rates as high as 18 per cent. More astonishing still, India has committed to purchasing $500 billion worth of American goods over five years. Which free-trade agreement contains a one-way obligation by one party to massively purchase goods from the other?

And this:

Acknowledging power asymmetry is realism. To internalise it so completely that one relinquishes independent judgement is something else. We might want to, in the name of pragmatism, give in; we could even make the best of this deal in a way that the consequences are not bad. But this is not a triumph. The perfume of official announcements cannot disguise the stench of our own diminishment.

A moderately ambitious India

Bhuvan — Mon, 09 Feb 2026 00:00:00 GMT

Pranay Kotasthane as usual with a thoughtful take on the state of affairs:

Lastly, the general impression I take away from this budget is the realisation of a party in power which has figured out a way to win elections in India that’s not contingent on strong economic performance. So long as you don’t screw up the economy badly, a combination of emotional issues and timely welfarism (direct cash transfer just before elections is a new, powerful tool now) should see you through most elections. Therefore, a 6.5 per cent growth and a 4 per cent fiscal deficit in a world that’s volatile and unpredictable is good enough to keep the global investors mildly interested while keeping the winning machine going domestically. That may not be enough for us to get out of the middle-income trap or lose out on the narrow window of demographic dividend we had. But the willingness to take bolder bets to address such structural challenges seems to have dissipated. More of the same is what we have settled for.

Be alone!

Bhuvan — Mon, 09 Feb 2026 00:00:00 GMT

This is music to the ears of an introvert and socially awkward guy like me:

“Be alone, that is the secret of invention; be alone, that is when ideas are born,” said Nikola Tesla in 1934. The inventor spawned his “earthly miracles” not in a busy lab, but in the countless hours spent tinkering away at electrical systems and radio technology on his own. “Originality thrives in seclusion,” he said, “free of outside influences beating upon us to cripple the creative mind.”

Philosopher Blaise Pascal made a similar point nearly 200 years earlier: “All of humanity’s problems stem from man’s inability to sit quietly in a room alone.” Even before our age of digital distraction, Pascal observed that people invent pleasures and conflicts like hunting, war and gossip to avoid the unease of self-reflection. But a bit of boredom, he insisted, can be groundbreaking.

Solitude is particularly undervalued in the workplace. While collaboration is important, decades of research on creativity have shown that working alone—even for 90 minutes—produces more original and higher quality ideas than group brainstorming sessions. The mind focuses and solutions emerge, unencumbered by groupthink and the pursuit of consensus often found in a conference room of colleagues.

Big tech capex numbers are ridiculous

Bhuvan — Mon, 09 Feb 2026 00:00:00 GMT

These big tech capex numbers are ridiculous

Big Tech companies will have to raise tens of billions of dollars to fund their skyrocketing investments in artificial intelligence this year, as capital spending outpaces cash flows even among some of the world’s most profitable companies.

Google’s parent Alphabet, Amazon and Meta all surprised investors with the scale of their AI spending plans over the past two weeks. A total of more than $660bn is set to be ploughed into chips and data centres this year as they race to dominate what many in Silicon Valley believe will be the biggest wave of innovation since the internet.

The unprecedented infrastructure build-out will force Big Tech executives to choose between stemming capital returns to shareholders, raiding their cash reserves or tapping the bond and equity markets more than previously planned, analysts say.

LLMs are bad at solving math problems not in their training data

Bhuvan — Mon, 09 Feb 2026 00:00:00 GMT

LLMs are still bad at math:

For the experiment, the authors — representing a diversity of mathematical fields — each contributed one test question that arose from research they had in the works but had not yet published. They also determined the answers; these solutions are encrypted online and will be released on Feb. 13.

“The goal here is to understand the limits — how far can A.I. go beyond its training data and the existing solutions it finds online?” said Dr. Kolda, who is one of few mathematicians to be elected a member of the National Academy of Engineering.

The team conducted preliminary tests on OpenAI’s ChatGPT-5.2 Pro and Google’s Gemini 3.0 Deep Think. When given one shot to produce the answer, the authors wrote, “the best publicly available A.I. systems struggle to answer many of our problems.”

Predictions are hard

Bhuvan — Mon, 09 Feb 2026 00:00:00 GMT

François Chollet on the money here:

Whenever I hear Very Serious Businessmen make confident pronouncements about the future of AI, I remember what the very same people were saying in 2021 about the Metaverse and NFTs.

When you lack a grounded, causal model of the world, your "predictions" are simply a remix of narratives you've heard from others.

Reminds me of something, actually

Good time to remember Bhuvan's law:

“Any discourse about artificial intelligence is indistinguishable from talking out of your ass.”

My adventures in vibe coding

Bhuvan — Sun, 08 Feb 2026 00:00:00 GMT

I wrote about all the side projects I've been building with the help of AI coding tools like Claude Code and Codex.

People were, are, and will be dumb

Bhuvan — Sun, 08 Feb 2026 00:00:00 GMT

Andy Masley here is quoting a tweet about the book The Secret:

A lot of people who think everyone went insane because of social media really don't remember how fundamentally insane people were before social media

A lot of otherwise normal teenagers and adults around me -Would buy bracelets with little crystals inside because the crystals provided energy -Thought 9/11 was an inside job -Didn't believe in evolution -Would talk about The Secret as if it were a respectable normal thing to believe in

I agree 💯

It reminds of this time when my personal life was volatile and this smart friend of mine recommended that I read this book. It may have been my lazines but I didn't end up reading it. But now that I think about it its wild how popular this idiotic book was. I've heard so many anecdotes about this dumb book from friends. In hindsight, it feels like the book was as popular as the bible at the time.

Look at the bullshit in it:

The second thing to know is that the condition of being overweight was created through your thought to it. To put it in the most basic terms, if someone is overweight, it came from thinking "fat thoughts," whether that person was aware of it or not. A person cannot think "thin thoughts" and be fat. It completely defies the law of attraction.

But to Andy's point, people always believed in dumb shit but it was private, but now everybody has the technology to broadcast their stupidity to the entire world at their fingertips. Now add LLMs into this mix. Brace for the wild times.

Introducing small web

Bhuvan — Sat, 07 Feb 2026 00:00:00 GMT

Small web started with a simple idea: build an aggregator for indie blogs and newsletters—a small corner of the web where you can discover personal writing and fall down interesting rabbit holes.

I think the initial spark came while reading about content slop online. I wanted to create an antidote to that—a place showcasing real people writing real things. With Claude Code, the backend came together in under an hour. I spent more time on the design, and what you see now is the result: a simple RSS aggregator that pulls from blogs and newsletters three times a day.

The selection criteria? Every single blog here is one I've read at least a post or two from over the years. No filler, no algorithmic suggestions—just an eclectic collection across various topics that I actually follow. I don't necessarily agree with everything, but they're all worth reading.

These days, I start most mornings here looking for my daily reads, and I'm often rewarded with wonderful discoveries.

I hope you find your next favorite read here too. Happy rabbit-holing. Please do share it with your peoples.

Kimi 2.5 is as good as Sonnet 4.5, ChatGPT O3 and Grok 4

Bhuvan — Fri, 06 Feb 2026 00:00:00 GMT

It's crazy how good good open source LLMs have gotten.

Kimi K2.5 set a new record among open-weight models on the Epoch Capabilities Index (ECI), which combines multiple benchmarks onto a single scale. Its score of 147 is about on par with o3, Grok 4, and Sonnet 4.5. It still lags the overall frontier.

Normie AGI is here

Bhuvan — Fri, 06 Feb 2026 00:00:00 GMT

Normie AGI Is Here

Picture this.

You have a tool at your disposal that knows everything about everything. It's trained on all of humanity's knowledge that's available on the internet, and it has internalized all of it. You can ask it about anything, at any time.

You can ask it whether the Medici family had a higher incidence of diarrhea in the 1500s. You can ask it if Baruch Spinoza would've been a brilliant shitposter. You can ask it if Karl Popper liked popcorn. You can ask it if Esha Deol is a better actor than Priyanka Chopra. (It'll say yes. It'll be wrong. But it'll say it with supreme confidence.) Or you can ask it to ghostwrite a LinkedIn dunk post, a thinly veiled rant about your boss's obsession with "synergy" and "circle back," so passive-aggressive it could win a Pulitzer.

It can also do things.

You can describe a product or an idea and it'll just... do it. It can build websites. It can build apps. It can analyze data. It can execute sequential workflows. You can take plain-English instructions and turn them into software artifacts. It can think, reason, and execute logical sequences of steps far better than you ever can. Its research capabilities are better than yours. It can search the entire corpus of humanity's digitized knowledge.

And it keeps getting better at a rapid clip. In fact, it keeps getting smarter at an exponential rate. A million zillion times faster than the rate at which you are getting smarter. Put another way: these LLMs are getting smarter at a rate faster than the rate at which you can swipe reels. Their increasing smartness is inversely proportional to the rate at which your brain is turning into mush.

Now, how would you describe this thing? This tool?

Wouldn't you say this is artificial general intelligence, however you conceptualize it? A superintelligent system? A supernatural system? A ghost? A wizard? A witch that can conjure things?

That's exactly my point.

I'm writing this on the day Claude Opus 4.6 and ChatGPT 5.3 came out. Actually, I'm not writing this. I'm voice-typing it into ChatGPT and editing it with Claude. Which, if you think about it, kind of proves the point. And there's no doubt in my mind that normie AGI is here.

Mind you: if you're reading this post, you are most likely a normie. Like me.

And if you're a normie who's deluded yourself into thinking these tools are useless? I hate to say it, but it's game over for you.

For all intents and purposes, the large language models as you see them, as you use them today, are artificial superintelligence for normies like you and me. At this point, if you have any doubt about whether these machines are intelligent, smart, capable, then it's as good as denying the fact that the Earth revolves around the sun... and that Ram Gopal Varma's Aag is so spectacularly bad that it's good.

The Discourse Is Missing the Point

It's February 2026, and if you spend any time online, people are still arguing: LLMs are useful, LLMs are useless. They keep asking which model is good, which tool to use.

Looking at these discussions, I can't help but feel like these people aren't going to make it.

Every morning, what's stunning to me is this: people have a disposable, intelligent tool trained on pretty much the entire digitized knowledge of humanity, and most people are oblivious. We don't have great data, but if you look at any usage or survey numbers, it's not more than 10–15% of people using these tools yet. Twitter and Bluesky discourse is not a good proxy for real life. Those are mostly savvy users who already get it.

Outside that bubble, you say "ChatGPT" and people will slap you in the face because they think ChatGPT is an abuse, like you abused their mother. Basically ChatGPT is a maa ki gaali. And this is... perplexing.

I constantly wake up and remind myself these tools are magical. If you went back to 1900, or even 2000, and described the capabilities of LLMs properly, the correct category for that tool would've been science fiction.

And today we're in this weird split reality. Some people are so used to these tools they think they're nothing special. Other people are oblivious these tools even exist. That dichotomy is just funny.

The fact that you have a tool at your disposal where you can prompt in plain English and it will produce output indistinguishable from a lot of experts... it shouldn't be real. It should be movies. It should be sci-fi. But it's our reality.

And yet so many people are laundering other people's opinions: "LLMs are useless," "they hallucinate," "they're stochastic parrots" (thanks, Emily Bender), "they mix things up," "they make mistakes."

To this I have to say: bro, have you looked at yourself in the mirror?

If you think LLM output is slop, don't make the mistake of thinking you're Michelangelo, da Vinci, or Einstein. You are sloppier than the sloppiest LLM. In fact, you're worse than GPT-2. That's what I'd say.

As a normie, the fact that I have access to these tools, where I can describe an idea or a site or a business or a hobby and they'll execute, still blows my mind. These were ideas floating around in my head. I had absolutely no business building any of them. I recently built a simple, beautiful RSS aggregator called smallweb.blog. Before that, I built a site that aggregates great poems in the public domain, with annotations and notes generated by LLMs to help people read and understand them, Poetic Reveries. I've built Akshara.ink where I'm trying to digitize Indian public domain texts.

It's ridiculous. I'm a normie. I don't know a lick of coding. I shouldn't be able to do this. Akshara is probably on the more complicated end of the spectrum when it comes to digitizing text. And the fact that somebody like me, with no tech jobs, no "linguistic background," no coding knowledge, can build something reasonably good and robust? It's kind of crazy. These things were not only within the realm of imagination, they're now within the realm of possibility, and I can't get over it. I've been heavily using these coding tools for a while now, and I still wake up surprised every day. That's saying something.

But the discourse is stuck on: "Is AI useful?" "Which tool?" "Is AI a bubble?" And sure, those topics have a place, but the way people obsess over them feels profoundly unhelpful. There are super-boosters and super-doomers, and even having a take on that seems useless to me.

I'm reminded of something Slavoj Žižek once referenced about quantum mechanics, Copenhagen school vibes: shut up and calculate. Don't get lost in the ontological metaphysical drama.

Today I'd say: shut up and use them.

It doesn't matter what you think of them. Until you use them, you don't get to have an opinion. And if you do have an opinion without using them, you're just another fucking idiot on the internet copying someone else's hot take and passing it off as your own.

I've seen smart people around me try these tools and get shocked by what they can do. Ideas they always wanted to execute, workflows they always wanted to build, the way they can change how they work. It's crazy. When I tell people what these tools are capable of, it doesn't land. I'm just another bloviating idiot praising AI. But when they actually use it, that's when it hits.

And by "use it," I don't mean the typical ChatGPT web interface. Those have their place. But until you use the coding tools, Claude Code, OpenAI Codex, Gemini CLI, Cursor, where you use software to solve problems and build things, you won't see the true magic. You won't see the normie AGI.

I highly recommend you do it.

Normie AGI arrived in 2024 itself, even before these coding tools became mainstream. But now it's normie AGI on steroids.

The Trade-offs

Now, I know how this reads. Another guy on the internet cheerleading for AI. So let me be clear: I'm not a booster.

Thomas Sowell said it best: "There are no solutions, there are only trade-offs." That holds for LLMs too. These tools aren't categorically good or bad. They have their uses, they have their place, and they have real costs. A few that keep me up at night:

Cognitive atrophy. You outsource your thinking, your thinking muscles weaken. That's just how brains work. If the LLM writes your first draft every time, what happens to your ability to stare at a blank page and think?

Judgment erosion. My whole argument about tacit knowledge being the last moat only works if you're still accumulating tacit knowledge. But if you defer to the LLM on every decision, when do you develop your own instincts? Heavy LLM use could be slowly draining the very moat you're counting on.

The illusion of competence. I can ship projects now. That's incredible. But building something is not the same as understanding it. There's a version of this where you produce a lot and learn nothing.

Taste convergence. Everyone using the same tools, trained on the same data, prompting in similar ways, risks converging on the same median output. The slop problem isn't just about bad writing. It's about a flattening of what gets made.

The questions I keep asking myself: How do you retain your sense of judgment when you're leaning this hard on LLMs? How do you decide what to outsource completely versus where you stay in the loop? How do you use the exoskeleton without your legs forgetting how to walk?

I don't have clean answers. That's sort of the point. And if you aren't holding two simultaneous, painful visions in your head right now, that these tools are magical and that they might be hollowing you out, if that duality doesn't create an annoying tingling or itching feeling in your brain's anal cavity, you're doing it wrong.

The Cope Check

I'm firmly of the belief that these large language models are better than humans at a vast majority of tasks and processes. Not outright replacement yet, but they are much, much better than us normies in a huge range of domains.

So where's the last moat? Human judgment. Knowing what to build. Taste. All the implicit, tacit, unknowable things you've accumulated through your lived experiences, those tiny little tidbits that aren't really on the internet, that luminiferous ether of knowledge floating around in your head. That stuff is not captured by LLMs. That's where we still have an edge. But only if you actually use that edge in combination with these tools.

And while I used to feel scared, now I feel liberated. Because I get to use them. I get to steer them. I get to wrangle them. I get to orchestrate them into doing things for me. This saves me time. It allows me to do more. It opens up possibilities I wouldn't have dreamed of. It's an exoskeleton.

Here's the meta point: I'm doing this sitting in front of a coffee shop. No laptop. No drafts. Just talking into a phone and publishing a blog post. That alone should tell you something about where we are.

If you, in 2026, as a normie, aren't intellectually honest with yourself, if you haven't taken an inventory of your capabilities versus those of large language models and arrived at the conclusion that they are fundamentally better than you at a vast variety of things, then you're just coping.

And if you're not using these tools, not following the AI discourse, not experimenting, not tweaking, not tinkering, not getting a sense of the shape of these capabilities, you are NGMI.

It's game over for you, bro.

Toni Morrison on evil

Bhuvan — Thu, 05 Feb 2026 00:00:00 GMT

Wow

“I was always a little bored by demonstrations of evil.

They relied on the same things. I used to say it always has this top hat and a cape and a tuxedo and a cane. And it has all this stuff. Glamour and costume. Maybe even a little theme music. But goodness never has anything, ‘cause it doesn't want anything. It can't use anything. It's just there.”

~ Toni Morrison (2013)

https://youtu.be/Dikga62kLYk?si=xWYzKmpbof-YcmIx

H/t Angus Hervey

Will Storr on smartphones

Bhuvan — Thu, 05 Feb 2026 00:00:00 GMT

Will Storr on what we've lost t smartphones:

Yet my smartphone has affected my life in often negative ways. The world I disappear to inside it has made me – and probably you – angrier. That’s my main impression of how the world has changed since 2007: we’re all a lot more pissed off with each other. And I really do blame phones. Humans are profoundly social and wired to solve the problems of existence by forming into collaborative groups. When we feel we belong and are valued, we’re happy; when we feel isolated and worthless, we become anxious and depressed.

Smartphones have gamified and monetised these powerful aspects of human nature. They don’t benignly offer us the connection and status we desire: they strategically withdraw it in order to drive engagement. Whenever we’re outraged by the behaviour of an identity group that’s not our own, it’s an attack on our status: we are drawn further into our phones to find out more and perhaps take part in a counterattack – an attempt to restore our threatened status and reinforce the connection with our team. We’re made to feel good or bad by likes, reposts, comments or follower-counts, but our phone issues these precious rewards unpredictably, just as a slot machine does – and just as Fogg described. It’s this unpredictability that helps make them compulsive.

159 years of momentum investing

Bhuvan — Wed, 04 Feb 2026 00:00:00 GMT

From Larry Swedroe's summary of this paper:

Momentum Works Globally The momentum effect appears consistently across international stock markets, with premiums comparable to those in the U.S—there were positive momentum returns in all 31 countries with sufficient data since 1990, with the majority showing statistically significant results.

Momentum Is Multi-Dimensional Perhaps the most innovative finding is that momentum extends far beyond simple price trends. The researchers identified and tested several alternative momentum signals:

• Fundamental momentum: Trends in earnings surprises and analyst recommendations.

• Firm-specific momentum: Returns after controlling for market, size, and value factors.

• Anchor-based momentum: Distance from 52-week highs.

• Network momentum: Trends among connected companies (by industry or analyst coverage).

• Factor momentum: Persistence in factor returns themselves.

Combining these various momentum measures into a multi-dimensional approach improved risk-adjusted returns compared to traditional price momentum alone.

Crash Risk Is Real, But Manageable The research confirms that momentum strategies are exposed to significant crash risk, with maximum drawdowns exceeding 88% during market reversals (particularly in 1929 and 2009).

Age of anxiety

Bhuvan — Wed, 04 Feb 2026 00:00:00 GMT

💯 We are headed for an age of anxiety, grievances, and disillusionment, and I don't think our systems, as they are designed, are equipped to handle the tsunami of angst:

THERE is still time for the cheerful and the gloomy to disagree, sincerely, about the net impact on employment from automation and artificial intelligence. Perhaps AI will unleash a job-creating economic boom for the ages, as boosters maintain. Maybe on balance the doom-mongers are right, and grim legions of algorithm-guided bots and robots will conquer one industry (and profession) after another.

One outcome is not up for debate. Across the industrialised world, wise politicians are braced for soaring public anxiety about change. Whether AI’s long-run effects are benign or cruel, many citizens are about to find once-valued skills and credentials surplus to the requirements of the economy. Even as new jobs are created, some flesh-and-blood workers may lack autonomy, or indeed dignity, if they feel they are serving AI masters.

Psychoanalysing Dario Amodei

Bhuvan — Tue, 03 Feb 2026 00:00:00 GMT

I havent yet read Dario Amodei's recent essay but jumped straight to this piece by John Herrman in which he reluctantly psychoanalyzes Dario:

It’s sort of hard to know how to read a manifesto like this from one of the most powerful figures in tech. Is it a sober, strategic precursor to policy papers for the next administration? The highest-profile episode of AI psychosis yet? A lament about the problems of today written in the technological dialect of tomorrow? If you take out the AI, it reads like a social-democratic electoral platform full of reforms and normative expectations that an American progressive would find appealing, resembling a plea to treat the tech industry’s future wealth accumulation as something akin to a Nordic sovereign-wealth fund. It’s likewise legible as a series of arguments about things that “we” should have started addressing a long time ago, like wealth inequality — partially a consequence of mass automations past — or the gradual construction of a terrifying surveillance state within a nominal democracy, with the help of the last generation of big tech companies. Amodei’s shoulds are, to his credit, more honest than the vague gestures at UBI or hyperabundance you get from some of his peers, but that also means they’re available to scrutinize. To the extent you can pick up on fear in “Adolescence,” it doesn’t seem to revolve around terrorists using AI to build “mirror life” that might destroy the planet or the prospect of that “country of geniuses” taking charge, but rather the way things already are and have been heading for years.

The future of work

Bhuvan — Tue, 03 Feb 2026 00:00:00 GMT

Martin Alderson on how AI is changing workflows in different organisations:

"This effectively leads to a situation where smaller company employees are able to be so much more productive than the equivalent at an enterprise. It often used to be that people at small companies really envied the resources & teams that their larger competitors had access to - but increasingly I think the pendulum is swinging the other way.

The future

I'm starting to get a feel for what the future of work looks like. The first observation is that (often) the real leaps are being made organically by employees, not from a top down AI strategy. Where I see the real productivity gains are small teams deciding to try and build an AI assisted workflow for a process, and as they are the ones that know that process inside out they can get very good results - unlike an often outsourced software engineering team who have absolutely zero experience doing the process that they are helping automate. I think this is the opposite of what most 'digital transformation' projects looked like in enterprise."

Coding is automated, vision is not