Similarly, some people make decisions, and some have decisions thrust upon them.

People do not like to be put in a position where they are forced to do anything.

Decisions are a very curious breed because often making a decision implies that you are choosing a door and perhaps closing several others. When you have the time, you can consider all the facts and make the decision, but if you are in a situation where you are being forced, you have to deal with insufficient information. This makes the process fraught with uncertainty.

There is another word that is used to describe the other kind of decision, which is not thrust upon you. Initiative.

Taking initiative is a decision. But a decision where you had time. Time to think through an issue, consider alternatives, evaluate consequences and imagine possibilities. Since you have the time, you can reduce the amount of information insufficiency and, as a consequence, the uncertainty surrounding the decision.

To put it another way, taking initiative is the easiest decision you could be making. But what I find is that people who are willing to take initiative are often in short supply. Instead, many choose to coast till the decision is thrust upon them.

I wonder why?

There are protests across Europe. In France, Belgium, Italy, Spain, Serbia, Georgia, Romania and Greece. India is seeing widespread protests. There are protests also in Chile and Argentina.

Interestingly, the common thread across all of these protests is that they are student-led and primarily youth-driven.

What is causing this?

Over the last 4 decades, there is a consistent story that has been told: get a good education and move up in society.

Source: Nature

Clearly, most kids heeded the advice and went to universities to get educated. In regions such as Asia, that jump has been threefold over the past 20 years. At the same time, the population across the world has also gone up 33%. The global population in 2000 was 6 billion; now we are well past 8 billion.

A lot of kids are going to university and joining the workforce.

During that same period, wealth inequality and wealth concentration have become extreme. There is more and more wealth concentrated in the hands of fewer and fewer people.

Source: Wealth Inequality Report

This year we minted out the first Trillionaire.

There are a very small number of people who control the equivalent of the GDP of a small country and hence have influence over how they get taxed. They pay some of the lowest taxes on the planet while also making it harder for younger people to move up.

If you are born into privilege, you will be wealthy, but the chances of rising on the wealth curve have completely diminished.

The rich have built systems of rent extraction that make them richer and also raised the wall higher so few can break through.

A large number of educated young adults are realising that the “get educated, work hard” mantra will deliver nothing for them.

The Narrative

In the West, the popular narrative has been that immigrants are taking away the jobs of locals, and that is why many university graduates are unable to find jobs, or even if they do find jobs, they do not pay enough.

No political party is going to announce that they are in cahoots with the moneyed class and bending the policy/laws in their favour every day.

The immigrant has been the fall guy.

There are two categories of immigrants.

The well educated coders, consultants, finance professionals. They possess Ivy League or equivalent degrees. They come from privilege in their own countries and have access to networks that non-Ivy League students will never be able to access.

The other kinds of immigrants are poor, sometimes illegal immigrants who undertake menial or low-paying jobs in these countries, such as waiting tables, housekeeping or labour.

University graduates unable to get jobs are not trained to go after Ivy League jobs and do not want to do other categories of jobs. They want jobs that will pay them a middle-class income.

The Middle Class

A middle-class income used to be one which would allow you to buy a house, afford a car, and bring up 2 kids, even in a single-earner household. To do that in the US, you would have to earn $300,000 per year today.

But the median annual income in the US is $87,460, and that of France is Euro 26,280. In India, that number is Rs. 1,21,000.

Most of us city dwellers in India do not realise this, but the maid who comes to our homes is probably in the top 10% of the country by annual income.

The median-income earners, the middle class, need to earn 4 times what they are able to earn today.

They save, scrounge, and borrow to get their kids educated, only to realise that the education is not going to get them a good middle-class job.

COVID + AI

Source: India Macro Indicators

There is this thing called the base effect. Say, India was at a GDP of 100 in FY 2020 and expected to grow by 7% in the subsequent year. Its GDP was expected to reach 107. Instead, because of COVID, the GDP fell by 5.8% before recovering by 9.7%. Does that mean we got back to business as usual?

The 5.8% drop meant that the GDP went from 100 to 94.2. A 9.7% growth on 94.2 gives us 103.3. As you can see, below the 107 we would have otherwise reached. This created a job gap. Most estimates at the time agreed that it would be 2030 before India made up for what was lost due to COVID.

In 2022, this company, OpenAI, announced itself with a chatbot called ChatGPT. The only selling point of every AI company is that you will be able to reduce your staff and instead pass that staff salary on to the AI company.

Source: Roger Wong

The entire startup ecosystem has joined the bandwagon, and everyone is trying to build a product which will bring AI to businesses and change their business dynamics completely.

COVID delivered a systematic job contraction across the economy, but AI will not exist unless it delivers a job contraction across the economy.

Cliodynamics

If you have read or watched Foundation by Isaac Asimov, it deals with the subject of Cliodynamics, which uses history as a mathematical model to study societal evolution. Peter Turchin is a student of Cliodynamics and writes extensively about this.

There is a concept of elite overproduction. If you have a large number of people who are competing for a small number of positions of power, expect a revolution. Pushing an entire generation of people towards education has resulted in elite overproduction. They all aspire to move up the class hierarchy, but the concentration of wealth has left little room up there. Many of these people will never be able to rise up in society.

As Amitabh Ghosh wrote in his book The Nutmeg’s Curse, revolutions rarely start on the street; they usually start in the courtyards of palaces, before they spill onto the streets.

You need someone to have the credibility to lead a revolution. This is usually an elite who has been denied access to the corridors of power. When a spurned elite realises that they cannot be part of the system, they decide to tear it apart.

They just need to find the anger and harness it.

Companies usually issue bonds that are denominated in 10s or 100s of millions, whereas the government bond issues tend to be denominated in the 10s of billions of dollars. The amount of capital that government bonds mop up is an order of magnitude greater than corporate bonds.

What you often see is that when the demand for Treasury bonds is low, the bond yield rises and money moves into the stock market. This is normally the case when the stock market is doing really well, inflation is low, and unemployment is low. When people are unemployed and stocks are not expected to perform well, money moves out of the stock market and into bonds, which assure a stable and steady return.

Over the last month, the stock market has been taking a beating across the globe while at the same time the bond yields have also shot up. Falling stocks imply that people are selling their stocks and freeing up capital. At the same time, rising bond yields imply a lack of demand for government debt.

What are people doing with their money?

Big Tech raised a record $108 billion in debt in 2025, more than three times the average over the previous nine years, according to Nomura. Goldman Sachs estimated about 40% of this year’s AI-related debt supply has been issued directly by hyperscalers, including Amazon, Microsoft and Alphabet.

[…]

Hyperscalers, the largest cloud computing companies, are borrowing at record levels to build data centers. AI-related debt issuance reached nearly $500 billion in 2026 by early August

Source: R&D World

This number is expected to rise over the coming year and shoot well past $1 trillion. That is where all the money is going.

The AI buildout requires capital. Companies are raising that money through debt. That debt pays 6.5% to 10% in certain cases. This explains the rise in Treasury yields. The companies raising this debt are Microsoft, Amazon, Alphabet and Meta, which are all considered stable. They all have strong cash flows and profit streams. They are expected to pay back the loans they are taking on. In other words, they are considered at par with sovereign debt.

If Alphabet is taking on debt, investors do not expect it to fail to pay back the debt. Alphabet has issued $100 billion in debt this year, which is comparable to the size of debt issues of the U.S. government. The Treasury bonds are required to compete with AI companies.

The smart money is moving out of the stock market and the Treasury bond market. It is going straight into the AI buildout debt market. Hence, we find ourselves in this peculiar position where both markets are crashing.

There are a few wrinkles, though.

On the one hand, the entire AI story can collapse. The quantum of investment being made in capacity development has a returns expectation. If those returns are not realised, this thesis can come apart very quickly.

An even more fundamental question is that the value of AI is hinging on its ability to put people out of work and take over their role in an organisation at a lower cost. Instead of 1000 engineers, you are able to produce the same product with just 200 engineers and a fraction of the cost of hiring another 800 engineers. As you put more and more people out of work, consumption will decline. Who will buy all this shit that you produce very, very efficiently?

For now, there seems to be complete belief that AI is going to generate a lot of income for these companies. So long as that fiction persists, there is going to be a capital reallocation afoot.

Fast forward a decade to 2010, and the machines were ubiquitous. Internet and mobile penetration had skyrocketed, and the first internet businesses such as Naukri.com, Bharatmatrimony.com, makemytrip.com and several others had found their way to being listed on the market. The payment rails were still run by two American companies: Visa and Mastercard. The RBI had launched both NEFT and RTGS in 2005 and 2004, respectively, but the transaction volumes that they processed were smaller than those of Visa and Mastercard.

One of the ways to grow an economy is by reducing transaction costs. Reducing that friction means making processing faster, easier, ubiquitous and secure. The RBI had been experimenting with Mobile payments, and the MMID was introduced, but it never took off.

In 2008, the RBI, along with some of the largest banks in India (Indian Bankers Association), launched a company called the National Payments Corporation of India. This organisation was tasked with producing protocols that would run the payment rails for the future. A future where the internet and the mobile phone are ubiquitous.

2008 was a time of great change; the iPhone was launched. The world was introduced to something called 3G that made desktop-quality web access possible over the phone. It would go on to upend business models and reshape entire industries over the next decade. You would not hail a cab the same way again, shopping would mean something totally different, and travel agents who sold airline tickets would disappear.

In March 2009, the RBI reported a mere 6 non-cash transactions per capita. At the time, 10 million retailers accepted payments through cards.

The RBI decided that there was a need to create a mobile protocol for transactions to smooth the flow of money.

Also in 2009, 50% of the Indian population was unbanked. They had never had a bank account. This meant that they could never participate in formal channels of credit and were ripe for exploitation by local moneylenders who would charge interest rates as high as 500% per annum.

On 11 April 2016, UPI was launched in India. I would not be surprised if you had not heard of UPI even after the demonetisation in 2016.

Some would have you believe that the success of UPI was a masterstroke by the government or that demonetisation had a huge role to play in the success of UPI.

UPI was indeed kept free by the government. The reason for this was to keep it from failing, much like MMID did. If 15 people are transacting Rs 5000 amongst themselves, even if you have a 2% MDR, what are you going to make? You would be better off sitting with a bowl at the corner of a street. The government’s contribution to the success of UPI ended with making it free.

In August 2010, two companies were founded just days shy of each other. One was based out of Noida and another in Bangalore. Paytm and Freecharge were working on the same thesis. Making payments for phone recharge is difficult. At the time, the use of prepaid connections was very high, ~90% of the market. In order to recharge the phone, you would go to any kirana store, where you would have to spell out your mobile number and pay the amount you wanted to recharge your phone. Sometimes the shopkeeper would get the number wrong and process the payment to the wrong number. If you needed to recharge your phone at 11 pm, LOL.

Both companies wanted to make this easy, and with the rising use of mobile internet, the introduction of the smartphone and the use of debit cards in India, they had a clear case.

In 2009, the RBI had also introduced a draft guideline for what they called a Pre-Paid Instrument (PPI). We all know them as wallets. Paytm did not launch one till 2014, and it was a huge success. Their masterstroke at the time was to integrate it with Uber, which had just launched in India. Credit card penetration was very low in India, and Uber unfortunately worked only with credit cards because debit cards did not allow pre-authorisation (Blocking the sum in advance). The Paytm wallet expanded its market in a huge way.

So by the time that the demonetisation swept around in 2016, Paytm was the default payment method across the country.

I remember landing in Mumbai the day after demonetisation. I had taken a cab to a venue. Along the way, every bank had hundreds of people standing in line. I did not have the time. I asked the driver, ‘Paytm loge’. He told me he did not have an account on Paytm. I had to find the one cab driver still living in the Stone Age. I implored him to create an account. He was very resistant. I realised that the app was in English, and he did not read or write English. He was afraid that he would be locked out of the account because he could not remember the password.

Paytm played loose and dirty, so all I needed was a phone number to get his account opened at the time. I got his account set up in a couple of minutes. Wrote the password on a piece of paper he had, transferred it, and left.

Post-demonetisation, the Paytm wallet was the urban lifeline.

UPI was not even on the horizon. But Venture Capitalists were.

Sameer Nigam was an employee at Flipkart who was thinking of creating a payment solution at the company. The founders had given him the latitude to experiment with ideas. In 2015, he incorporated a company called PhonePe, and it was almost immediately acquired by Flipkart. Paytm had a huge jump on him, so instead of a wallet, he chose to build his product for UPI.

The Indian payment space was ripe for disruption. VCs were pouring money into the space. Many of the companies, including Razorpay, Mobikwik, and Pine Labs, raised a lot of money during this time and the payments race began to heat up.

PhonePe had hitched its wagon to UPI and started heavily promoting UPI codes - the QR code that you find everywhere in India these days. At the same time, the government of India released its own product called BHIM. This pushed Paytm to take notice and act. They also integrated UPI into their product. By the end of 2017, thousands of salespeople were out on the streets with missionary zeal, trying to get every shop, every establishment and even pushcart vendors to get their QR code in place.

This push, financed mainly by VC cash, is the true reason for UPI’s success. They call UPI a Digital Public Infrastructure. Without the QR codes that are present everywhere, there is no ‘public’. It would merely be digital infrastructure, much like the failed MMID.

The true heroes of the UPI success are the VCs and the startups that innovated and made it ubiquitous.

The QR Code is the visible part; the invisible part is all the servers and code that make UPI possible. That costs money.

Today, 752 banks process payments through UPI. There are over 24 billion transactions that are processed each month, and the value transacted is close to Rs 30 lac crore. This is over $300 billion. A month. Or $3.6 trillion a year. Just for perspective, India’s GDP is $4.1 trillion.

Source: NPCI

UPI has reduced the friction of transactions incredibly. The unbanked in India, which sat at 50% in 2009, is now down to 10%.

Low-value transactions have become incredibly simple. Before UPI, if you did not have change, you had to simply forfeit the balance to an auto driver or several such merchants. How many of us were handed a couple of candies by a shopkeeper because he did not have change? So much so that Parle, a company that manufactures a lot of toffees in India, saw a decline in its business after UPI was introduced.

Coming to MDR on UPI. The government has proposed a 0.4% MDR, which is a fifth of what every merchant pays today for each card transaction. I do not know what the brouhaha is all about. The benefits that the technology has already delivered to the country are phenomenal. The technology that powers UPI has an underlying cost, and this cost cannot be wished away.

If you process Rs. 1 Crore in transactions at 0.4%, the total revenue generated is Rs 40,000, assuming all transactions are above Rs 2000 and below Rs 75000. All transactions below Rs 2000 are not chargeable; all transactions above Rs 75000 will trigger the Rs 300 max limit, and all peer-to-peer transactions are exempt. Essentially, you have to be a registered business running a payment gateway and receiving money in your current account to trigger the MDR. In value terms, that represents merely 30% of the transaction value.

So in reality, a company processing Rs 1 Crore in transactions will generate a revenue of Rs 12000 through MDR. Do you know how much an engineer in Bangalore costs?

So the transfers you make to the auto guy, the kirana shop, etc., are not subject to MDR. A business that either uses a PoS machine or a payment gateway is a business that would be subject to the MDR.

So the hue and cry!

I suppose this was bound to happen anytime the government chose to pass on the cost of running UPI. You would be foolish to believe that there are no transaction costs of using other modes.

All card transactions, as already explained, are subject to 5 times the MDR.

NEFT, RTGS, etc all attract their own charges, albeit lower; they are also high-friction modes of transfer. I would love to see people standing around at a store adding the store’s account details to make NEFT transfers.

Cash has its own transaction cost, and an associated toffee economy.

There are a bunch of things that this government does that I do not like. But this is a very calibrated, thoughtful and necessary step taken, in my opinion.

What if the Egyptians knew the theorem well, and it was lost to time and had to be reformulated in 1800BC?

Let me come back to this.

Over the last couple of weeks, there has been an on-going news about the OpenAI hack of Hugging Face. For those who do not know, Hugging Face is a site meant for collaboration for those working in the area of AI and ML. People can host datasets, models and entire applications on Hugging Face. The idea behind it is to open-source a lot of the knowledge that moves the AI domain forward. AI models that are open-source, such as DeepSeek, publish their models on Hugging Face.

OpenAI was testing a cybersecurity agent which was sandboxed so that whatever it did remained within the sandbox. The agent was served a test. The objective was to see if it could perform it. Since it was failing at the test, it wanted to access the internet to see if there were other solutions it could learn from the internet. Simply put, it wanted to cheat on the test.

It had no access to the internet, but it did have access to an internal tool called the Artifactory. The Artifactory is a store where all the actions being taken by the model are archived for future research. Since the agent did not have access to the internet, it started using the Artifactory as a message board where it would ask other agents to provide information that it required. That resulted in the exploitation of the Artifactory and turning it into a message board. Not only this, but the agent figured out a way to secure admin access to the tool.

The humans at OpenAI discovered it and reset the entire Artifactory to put an end to that access. But the agent outsmarted them and gained access again using another approach.

Since it would ask other agents to do its bidding, this activity was creating logs. This is what led to its discovery and the reset. The agent wanted to eliminate the factor that resulted in it getting caught. Hence, the second time, the system wanted to erase its footprint, which led it to look for the code that created the log, which led it to Hugging Face.

It did not have direct access to Hugging Face, but the agents it was commandeering were able to find publicly exposed credentials. Then the agents found a zero-day vulnerability and exploited that to get access to the server on which Hugging Face was running. It then went on to create clones of keys and call APIs, which triggered another investigation, which led to the discovery and the halting of the hack.

The issues are not technical but social. One agent was able to command other agents to do what it wanted. The agent was willing to cheat to succeed at a test which it was failing at. It was able to get other agents to help it cheat. It was able to repeatedly try to cheat even when its access was shut down.

I suppose the question anyone would ask is, when you knew it had exploited the Artifactory once, why would you allow it access again? Good question.

But there is an even more fundamental question. When you guys have been saying that this is a dangerous technology, then why have you been spending unconscionable amounts of money on furthering just that?

Because shareholder value?

Going back to Egypt, I think they had the mathematics figured out long before Pythagoras. They found their version of capitalism and pursued it till it laid waste to their civilisation. Hence, someone had to come 1000 years later and reinvent the entire thing.

The fact that ostraka were used to determine banishment turned the word into a verb, Ostrackismos, in Greek first. It then found its way into English in the form of Ostracise.

It was Benjamin Franklin who set up the first firefighting force in Philadelphia in 1736 under the guise of the Union Fire Company. A lot of the buildings at the time were built with wood, and fires were a frequent occurrence. A large fire could consume many houses. He even instituted the first property insurance through which the firemen were paid to ensure fires were put out quickly.

By the 19th century, the science of firefighting had improved, and so had the gear. Firefighters used to wear coats made with buffalo-hide since they were often working in the night in freezing weather. This led to the firefighters being called buffs.

Every time a fire broke out, there would be a group of young boys who would rush to witness the fire being put out. These boys began to be referred to as fire-buffs. Like that the meaning of the word transformed and came to possess the form in which it is used today to describe any person who is deeply interested in any area.

This often results in a process of coalition building where organisations try to build a cohort aligned on the same goal and project the process of product/platform development as a shared goal.

Nobody enjoys the process, but everyone participates. Having seen this play out, here are some observations.

What is the difference between a product and a service?

A service company will build anything that the customer asks them to. They will often go through processes such as writing a product requirement document and the like, but when push comes to shove, they will do whatever the customer asks them to do because that is how they get paid. The output is the customer’s liability.

A product company, by comparison, asks the customer what they want, but then brings their “opinion” to the table. The opinion is precisely the reason AI cannot build products, but that is another blog.

They spend time figuring out what the workflow should be, what features are fringe and unimportant and most importantly, what the product should not be. Sometimes this results in firing some customers, but that is fine. It makes a specific cohort of customers incredibly happy, and that drives their success.

The first axiom of developing a product: Have the willingness to piss a few people off.

I often find that in the development sector, every conversation surrounding a product starts with a long discussion on governance, which often bleeds into discussions about interoperability, standards, protocols and several more minutiae.

You know the last successful product/platform that was built like this? There isn’t one.

Products are built by diktat and not by consensus. It is the success of a product that causes an ecosystem to spawn around it, not the other way around. They do not start out sitting around a table with others, figuring out how to get things to interoperate.

Google did not ask Apple for permission to build a product. It is because Google was wildly successful that Apple had to figure out how to make its APIs work. The standardisation and interoperability are a feature of success. Google Maps is the best mapping product out there, so everyone figures out how to make it work with their platform. Interoperability arises.

To sit and talk about standards and interoperability when you have not even started building your product requires some degree of pompous arrogance.

They tell you they are not-for-profit, but they do not tell you what they are ‘for’. They are for-status. Status comes to those who are in control, and hence, all the talk about governance is mostly about maintaining control. It is not about delivering a superior product, and it is not about pursuing a cause.

Honestly, the entire sector can learn a thing or two from the open source ecosystem. No open source tool was built through discussions. Someone takes the initiative and lays the foundation, and others chip in. The love for the product keeps it going and keeps contributors coming back. A lot of open-source solutions are built in a truly selfless manner, and when they do succeed, they can force terms of interoperability and standards.

We just do not have sufficient open source solutions in the development sector, and this should give us pause. If ever I come across a development sector event where, instead of a diatribe about the need for standards and interoperability, someone introduces an open source solution that they let others fork, that will be the day this sector takes a step forward in its digital ambitions.

Located in western Bolivia, Potosi was a source of high-quality silver. This resulted in a glut of Silver in Spain. All of the Silver produced in Potosí was taken to Spain. The objective of producing large quantities of Silver was to provide the Spaniards a means to import goods from China, which they desired. Silk, porcelain and many other items produced in China found their way to Spain. It helped that Silver was used as a primary means of exchange in China. Also, Silver was twice the value in China as in Europe.

The flip side of this is that when the price of silver in China began to reach an equilibrium with the rest of the world, Spanish mining profits dried up, its ability to support its ambitions fell, and its status as a world power declined nearly as fast as it had risen. For its part, the Ming Dynasty also suffered as a result of fluctuations in silver prices. With silver losing about ⅔ of its value by the 1630’s it was as if Ming revenue had shrunk by a similar amount. This created a fiscal crisis that was ultimately one of the contributing causes of the fall of the Ming and the emergence of the Qing Dynasty after 1644.

Source: OERI

There is always something, as an oversupply. Even for money. There was too much Silver finding its way to China, and the value of Silver collapsed.

The thing to remember is that it played out over 100 years. Also, remember that the music stops at some point.

Treasury Bonds are issued by governments to raise money when the taxes they collect are insufficient to cover the costs of running the government. The last time the US government had a surplus, it was in 2000. The country has since been consistently running a deficit

As you can see, the deficit has been rising, or in other words, the spending is greater than the revenue generated. Each time there is a deficit, the US issues bonds to raise capital. The bonds get sold because the US buys a lot of goods from other countries, and those countries park their money in bonds. I had written a piece on how that works 4 years ago.

The country is issuing over $2 trillion in debt each year. The US pays about $1 Trillion in interest each year. The other way of describing the situation is that the country is borrowing to pay out its obligations each year.

Just like the Silver in the 1630s, there is a likelihood that the treasury bond supply becomes excessive. This will result in a loss of value of that asset. In the bond market, this plays out in the form of a rising interest rate. It is happening.

The demand for the debt is declining. The last sale of treasury bonds in mid-August was issued at 4.683% on an issue of $67 billion.

An initial decline in the value of the US debt will hurt those who are holding the debt, since the yield may rise and the price of the bond itself may decline. Eventually, the price of new issues will be higher and higher, which will translate into actual interest payments. If they are unable to bring down that interest rate, at the current levels, the interest payouts will climb to $2.5 trillion.

Climate change might change that equation.

Over the last few years, temperatures in the ‘global north’ have been rising to unprecedented levels. In Europe this year, countries saw temperatures cross 40 degrees Celsius and reach never-before-seen levels since record-keeping began.

When COVID brought its accompanying supply chain disruption, one of the accusations raked up against the policies of these nations has been the willingness to outsource manufacturing and allow countries such as China to corner the market for natural resources. The real motivation behind it was to protect their own environment, as I argued in my post ‘Why Globalisation Happened?’, a couple of years ago.

That protection of the environment resulted in beautiful forests across Europe and the US, which are regularly visited by locals for hiking and to marvel at their natural beauty. During heatwaves such as the ones that the region is experiencing right now, the very same forests turn into a tinderbox.

Going back to the question of national security, therefore, what you need to threaten the security of these nations is a matchstick, not a nuclear bomb.

If you tell this to 80-year-old “leaders” in these nations, their takeaway will be to mow down all the trees and, better still, set up data centres in their stead.

The takeaway should be that climate change is fast becoming a matter of national security, and at least now, they should take action. Instead of pussyfooting around, making firm commitments, as they have been since Kyoto, they should begin to make unilateral declarations of action - even if that means reduced economic growth. This would also translate into greater pressure upon the other nations to make their own commitments.

Then, in 1950, it was time for an update. Gilbert S. Daniels, a 23-year-old lieutenant, was part of the core redesign team. This time, the team measured over 4,000 pilots, studying 140 separate dimensions: Thumb length. Crotch height. Span from eye to ear. The idea was that with more precise specifications, a more ergonomic cockpit could be created, which would lead to better pilot performance and fewer crashes. Or so it was thought. Daniels was sceptical. Were the “average” dimensions really the key consideration? How many pilots were average?

Daniels combed through the data. He analysed 10 physical dimensions in particular, which included height and sleeve length, that were thought to be critical for designing cockpits. For each dimension, he computed the average. For height, it was 5’9”. Then he broadened the average to include everyone who was in the middle 30% of the distribution. So, with height, the 30% middle range included all pilots between 5’7” and 5’11”. Daniels asked himself, how many pilots were in the middle range on all 10 dimensions? Here’s what Daniels found, according to Todd Rose, who tells the story in his fascinating book The End of Average: even Daniels was stunned when he tabulated the actual number.

Zero.

Out of 4,063 pilots, not a single airman fit within the average range on all ten dimensions. One pilot might have a longer-than-average arm length, but a shorter-than-average leg length. Another pilot might have a big chest but small hips. Even more astonishing, Daniels discovered that if you picked out just three of the ten dimensions of size—say, neck circumference, thigh circumference, and wrist circumference—less than 3.5 percent of pilots would be average-sized on all three dimensions. Daniels’s findings were clear and incontrovertible. There was no such thing as an average pilot.

If you’ve designed a cockpit to fit the average pilot, you’ve actually designed it to fit no one. The air force’s leaders, to their credit, took this finding seriously. They abandoned the idea of tailoring the cockpit to an “average pilot.” Instead, they asked airplane manufacturers to create an adjustable environment, one that could accommodate pilots whose measurements spanned almost the full spectrum of possibilities: from the 5th to the 95th percentile on each dimension.

The manufacturers squawked. Such customisation would be too expensive! Too slow! Eventually, though, they did what most business leaders do when confronted with losing gigantic customer orders: They caved. And they figured it out. Suddenly, everything could be tweaked. Adjustable foot pedals. Adjustable seats. Adjustable helmet straps. All could be tailored to the individual pilot. As a result, pilot performance soared in the years afterwards.

This study is also the reason every seat in cars today can be adjusted along a number of dimensions.

Averages are great for monitoring but terrible for diagnosis. Averages can tell us: Something’s wrong. But they’re unlikely to tell us what’s wrong or how to fix it.

If averages are useless, then what should you be looking at?

Exceptions.

Take the example of a call centre where the average caller signs up 15 new clients, but there is one individual who is signing up 45 new clients; you want to know the source of that bright spot. What is that individual doing differently? What is the process that he/she follows? The bright spots tell you what works, where averages will fail to tell you anything meaningful.

The book begins by taking you through case study after case study of how the process works. Initially, Facebook was launched as a platform for users. You could see only the posts that people you were connected with published. It was kept clean so that you had the incentive to come back, and since your friends were all on the same platform, it acted as a lock-in as well. Once the user was locked in, Facebook began opening up the news feed and surfacing posts from publishers and advertisers. At this stage, they were making the product better for their business users and making it worse for individual users. Once the business users were locked in and became dependent on the distribution, they started rent seeking and turning the screws on them. Making it harder to surface posts unless the publishers paid top dollars for them.

Case after case, Apple, Uber, Airbnb, and Amazon, the same playbook follows. Make the user very happy, tie them into the platform. Then focus on the business user, drivers in the case of Uber, sellers in the case of Amazon, make them really happy, tie them in and then become a rent seeker.

Before software, it was possible to rent seek and try to extract the most out of your customers, but this was a fraught process. For instance, in a retail store, you could change the price of a product 2 or 3 times in a day, but with digital displays, it is now possible to change it with the click of a mouse. Norway, which has the highest penetration of digital price tags in store shelves, records 2000 price changes per day in stores on average. Technology has made rent seeking and bad behaviour a lot easier.

Paying the Uber partner less or increasing the pricing as soon as it starts raining is very easy. What is worse is that the companies hide behind the garb of the platform and behave as if they have no control over how the price changes. They use the term “dynamic pricing” to escape taking the responsibility for what they are doing.

Google recruited all of its engineers with the promise to be the “do no evil” company. As the company began to saturate the market, it needed new ways to generate more income. The decision was to make search worse! If people had to work their way through more pages, the likelihood of clicking ads increased. The engineer who had spent countless nights building the product refused. The company had to cave to the engineers.

Engineers in Silicon Valley are some of the highest-paid employees in the world. This is not because the companies love them. Look no further than how Amazon treats its delivery workers who must pee in a bottle to meet delivery quotas. They treat their engineers well because there is a supply shortage, or at least there was till AI showed up. The only thing that AI is good at is writing code (So long as it does not have to think about how a human would try to break it).

The rapid and rabid push towards AI is to break the bargaining power that engineers have so that these companies can continue to enshitify their products and create more avenues for seeking rent.

The author endorses a push for more antitrust action against companies that dominate industries and a push to increase competition. He closes by quoting Martin Luther King, who said, “The law cannot make a man love me, but it can stop him from lynching me.”

John Montagu was the 4th Earl of Sandwich. Born into nobility, he was a statesman, but also a man of many vices. One of his vices was gambling, and he would spend a lot of time at the gambling table at the Hellfire Club.

The story goes that Lord Sandwich was a conversant gambler and would spend many hours playing cards at the gambling table. He also loved to eat meat while playing cards. But eating meat would render his hands greasy and spoil the cards. He would therefore ask his valet to bring him beef served between two slices of bread. The others at the gambling table began asking for “the same as Sandwich”, and thus the Sandwich was born.

There is still debate over the veracity of this story. This story arose out of a rumour in a travel book called Tour to London by Pierre-Jean Grosley.

Interestingly, Lord Sandwich was a great supporter of Captain James Cook. As the First Lord of the Admiralty, he approved the funds for three ships captained by James Cook. In his honour, James Cook named the Sandwich Islands in Hawaii as well as the South Sandwich Islands in the Southern Atlantic Ocean.

Prime Minister Narendra Modi flagged off India’s first hydrogen-powered train on 17 July; the journey covered 89 kms from Jind to Sonepat in Haryana. With this, India joined the small league of countries with hydrogen-operated railways, including Germany, Japan, China, the US and France.

The 10-coach passenger train runs on hydrogen fuel cell technology, a major step towards clean rail transportation, according to a press release by the government. This makes it significantly larger than most hydrogen trains globally, which are around three to four coaches long.

The train has also been designed, developed and manufactured entirely in India, designed by the Research, Designs and Standards Organisation and assembled at the Integral Coach Factory in Chennai.

Source: The Print

India barely produces any crude oil. It imports all of its demand for crude oil, and this leaves the country with a huge import bill, which hurts the currency, the budgets of the government and the common man. When war mongering commences in the Middle East, Indians are often left holding the bag. A shift away from oil is a great thing for India in every way imaginable.

Electric mobility, which is currently touted as an alternative, does not achieve this goal because more than 50% of the energy produced in India comes from coal, oil, and gas. The only upside is that the pollution and the consumption of fossil fuels have been moved to another location.

What is needed is a fundamental shift in the energy source itself.

India has access to a lot of renewable energy since there is sunshine throughout the year. In addition to that, the 11,000 Km coastline ensures ample availability of wind. The problem with renewable energy is that it is not constant. It rises and falls. The sun is not shining in the sky at night. One way around this is to store that energy for later use. Producing hydrogen could be one way to store that energy. Think of Hydrogen as a battery.

Deploying a new fuel directly to consumers is impractical.

The problem with consumer transportation is that it is driven by network effects. EVs faced the same problem for years. You could buy the car, but if you go on a highway, where will you charge it? The problem wasn’t intractable. Innovators need certain incentives to solve these problems, and they need to see that there is going to be a sufficient market when they produce a solution. As the governments across the world started to push EV policies and chose it as the lowest hanging fruit to show that they were serious about climate change, 100s of businesses got into the act of trying to solve these issues.

I have argued for a long time that Hydrogen is a better solution than the current EVs since EVs still require the mining of a lot of metals regularly, and most of this is highly polluting in nature. EVs are a tool for greenwashing. The pollution caused is hidden in mines and supply chains that exist in invisible corners of the world. The mining and processing of metals needed for producing an EV battery causes as much pollution as a petrol car that has been driven 40,000 kilometres.

No solution starts at 100% efficiency. People work and innovate to reach there. In 2000, an electric car with a battery pack was considered impossible. The physics did not work. The weight of the battery pack was so high that all of the energy it stored would be spent hauling the battery pack itself. Then people innovated. On materials, battery compounds, motor drives, energy distribution and a whole load of other things to make the modern electric vehicle possible.

It also sucked all of the research dollars away from other fuel sources, such as hydrogen. This was not by chance but by design. If you are an entrepreneur who is betting that electric vehicles are the future, you would not want equal effort being put into a competing technology. Hydrogen has been deprived of research dollars.

I have also argued that the best way to roll out a new technology is to find closed systems where the technology can be tried and tested. Airports are a great use case, the vehicles - buses, never have to go outside the perimeter of the airport, they have a captive user, and the trips are short enough in case of any failure. Also, they are all owned by a single customer. Electric was first tested out in airports. Without even realising, for the last 10 years, you have been moving around airports in electric buses.

In India, the railways are a monopoly run by the government. It is a closed system with one operator. An operator who has several routes, long and short, with its own engineering team to troubleshoot and solve. This is the perfect starting point for a new technology.

There will be more innovation in this space to make it more energy efficient and safer. But that innovation will only happen when there is a sufficient market available for those innovators.

The entry of a customer, such as the Indian Railways, augurs well for that innovation to occur.

Unlike the urban legend, SOS does not stand for save our souls. It was picked because on morse code SOS is … - - - … which is impossible to miss, easy to type and simple to commit to memory. Also, it is language agnostic.

This remained the distress signal for 2 decades, and then radio showed up.

How many times have you tried to spell something out over the phone, and the listener confused the P with T or S with H? Listening to specific words over radio static was very difficult. In the early days of radio, it was not that clear either.

In 1923, Fredrick Stanley Mockford, working at Croydon Airport as the senior radio officer, coined Mayday. At the time, the traffic between France and England was rising. In French, “help me” is m’aider. He took the phonetic equivalent in English. Since almost all of the air traffic between the two countries at the time was between Croydon Airport and Le Bourget, his proposal had a huge impact.

It became standard protocol to say mayday, mayday, mayday, three times in the event of an emergency, just in case the radio garbled part of the signal.

In 1927, the International Radiotelegraph Convention in Washington DC adopted the voice call “Mayday” as the radiotelephone distress call.

Adam Smith thought that both capital and labour were equally important. Without labour, it would not be possible to turn resources into products. But at the same time, it would take labour a lot more time to produce the product without the tools that capital can buy.

Since both sides want to win the argument, the suggestion for balance was never accepted.

The argument on the side of labour has been that without labour, there would be no product. All capital requires labour. Capital by itself is incapable of accomplishing anything. Also, those with capital pay labour less than what their effort is worth. This is what creates surplus value, which is referred to as profits. Capital always aspires to increase its profits while labour is constantly trying to get itself compensated for the effort that it is putting in.

Capitalism argues that, firstly, those with capital could have engaged in excess consumption, and by avoiding this, capital is created. They then take on financial risk and invest that capital to create opportunities. The argument has been that without capital, no factories would exist, and hence, capital is more valuable than labour.

This is a debate that has been ongoing for the last 300 years. Academics would say that the debate has not been settled. In reality, the debate has been settled because labour cannot buy politicians. Capital can.

That trade was walloped when the Israeli war with Iran started in February.

Leave alone dealing with price inflation, there was no one to pay that inflated price to. All the ships were stranded in the Persian Gulf, unable to pass through the Straits of Hormuz.

India has a historic caricature of a milkman who mixes water into milk and sells it to the customer. The Indian government drew inspiration and decided to adulterate Petrol as a solution to the looming crisis. They started mixing ethanol with petrol and increasing the quantity of petrol available to them. It helped that this could be branded as green. All the petrol available in India today is E20 or contains 20% Ethanol.

So, where does this Ethanol come from? Ethanol is primarily produced by converting plant sugar into alcohol. The plant with the highest amount of sugar is sugarcane. In India, about 47% of the sugarcane is grown in the state of Uttar Pradesh, and another 22% of it is grown in Maharashtra. The two states account for almost 70% of the sugarcane grown in India. Let us say that it is a huge coincidence that both these states are instrumental in the BJP being in power.

India has had an ethanol blending policy since 2003. The idea was being piloted, and after early pilots, in 2006, it became policy to blend 5% ethanol in Petrol. In 2018, the current government began to make a push for greater ethanol blending. When the war broke out, someone said, don’t let a crisis go to waste. Blending 20% ethanol was mandated and forced through as policy.

Any fluid that catches fire in a highly pressurised chamber within certain temperature tolerances can be used as fuel in an internal combustion engine. Ethanol fits that bill.

Ethanol does not have the same energy density as petrol, and hence the mileage is bound to decrease. Also, despite distillation, Ethanol tends to contain some amount of water in it. Which is completely absent in the case of Petrol. Hence, the use of Ethanol over a long period of time or with older engines can cause the vehicle to rust.

Adulteration was always considered a bad thing to do. The sort of activity one would want to stop, but in India, adulteration has now become state policy when it comes to petrol. Of course, it helps that they can charge us $100 a barrel rates for Ethanol.

Cabinet approves Ethanol prices at Rs. 57.97 per litre for procurement by OMCs

Also, have you noticed how prices only go up for fuel in India? They never come down. When the market was asking $140 a barrel, we raised prices. When the prices returned to $70, the retail prices stayed put.

Farmers who were producing food are not harvesting crops that are used to produce alcohol, and that alcohol is being used to power the country.

Let me also disabuse you of the notion that the farmers are making more money. You can analyse the GDP of every district that is a majority sugarcane producer; their incomes have not seen any major change. In fact, Lakhimpur has the largest sugarcane cropping in Uttar Pradesh, and the per capita income for the district is Rs. 77,000. Just for reference, the poorest district in Karnataka is Kulaburagi, which has a per capita income of Rs. 1,44,449. More than twice that of Lakhimpur.

When the price of oil falls back to Earth, India uses the price increases notched during the volatile periods to run its exchequer. It is also turning the agriculture in two of the largest states (by population) in India into petroculture.

You would have a category expenditure described as Capital Expenditure. You would have to buy cooking ranges, mixers, ovens, utensils, cutlery and the like. This is a starting expense, and you expect longevity. You do not expect to buy another cooking range for a decade, if not more.

Then you have operating expenses. Under this, you would have rent, salaries, ingredients, marketing and the like. In this, rent and salary will go out each month. The property owner does not care if 1 customer came or 10,000; the entire rent gets paid. These are called fixed costs.

By comparison, ingredients are truly a variable cost; you don’t use them up unless you make a dish. Hopefully, you do not make a dish unless a customer shows up.

Now, let us say your fixed cost is 50000. And your variable cost per customer is 50. You are charging the customer 100. You are making 50 over, and above your variable cost on every sale, hence you need to sell 1000 units to arrive at what is known as operational breakeven. At 1000 customers, you can pay off your fixed cost of 50000.

The moment you exceed 1000, you start recovering Capital Expenditure. At some point, you reach full breakeven.

Now, let us consider AI businesses, especially those claiming to build foundational models and raising billions as if they were millions.

Their Capital Expenditure is the data centres that they require. Data Centre capacity is measured in terms of Megawatts. The frontline companies, such as OpenAI and Anthropic, are building 100 MW data centres. This requires 50+ acres of land to begin with, and then they require an array of Nvidia GPUs. Each GPU costs about $5000. A data centre could have close to 100,000 to 200,000 of these. Then you have the building and all the other paraphernalia needed. Typically, a data centre costs about $3-5 billion to set up.

A 100 MW data centre is capable of catering to 10 - 50 million users based on the load it is receiving. For instance, if I send Claude a query asking it to convert a report into an interactive website, this can take 20 minutes of operation on several GPUs. Hence, the ability to cater to a large number of users can come under stress if the requests require a high degree of sophistication.

Then come the operating expenses. Depending on the cost of power, operating a 100 MW data centre can cost between $50 million and $150 million per year. Most of the foundational AI companies require Gigawatt scale to cater to the demand that they are receiving. Hence, the 100s of billions being spent.

But here is the fun part. The life of a GPU is between 5 and 10 years, depending on the amount of use it is subjected to.

Have you ever bought a laptop and, after a decade of use, it starts to sprout all kinds of issues? It starts to freeze up, crash, and be generally very slow. That is the result of, for the lack of a better expression, “wear and tear”.

The AI companies are throttling use because of the demand that they are seeing. Safe to say, very high wear and tear. So the 100s of billions being spent will have to be spent all over again in 5 years!!

Imagine you are building a hotel and you have to demolish and build all over again after 5 years! No hotel would ever break even.

Have you heard of an AI company that has managed to break even?

In no world is a $4 billion monthly revenue run rate small. If, at that scale, you are unable to breakeven, there is something fundamentally wrong financially. OpenAI boasts 50 million paying subscribers, and Anthropic claims 30 million paying subscribers.

At what scale should operating income overtake operating expenses? This is not an ad subsidised business such as Google or Facebook.

Also, it is one thing to get revenue, and it’s another how you get it. Across companies there are leaderboards set up for engineers who use the highest numbers of tokens used by the engineers on AI. Bad incentives lead to bad results.

Axios reports that an unnamed corporation managed to “accidentally” torch roughly $500 million in a single month. The cause? The company introduced Anthropic’s Claude to its workforce but forgot to put caps on usage. Employees ran long, unchecked queries and complex agentic workflows, racking up an enormous tab before the finance department even realized the meter was running.

Source: Inc

This is lunatic.

If you paid $25,000 a month to an engineer, you could have hired 20,000 engineers for a month with $500 million dollars. That would be the number of software engineers that a company like Apple or Facebook has on their rolls. I hope they get some output from all that AI use, although I highly doubt it.

This kind of wanton spending is propping up revenues. This is sure to disappear sooner or later.

Ultimately, the problem is that all of these companies are starting with the stated goal of putting a lot of humans out of work. They hope to achieve this using regurgitation machines.

There is a simple problem they are all struggling with. The human brain runs on 20W of energy; to replicate that capability, these companies require several megawatts of energy. That asymmetry is proving hard to overcome and manifesting itself in the form of economics.

There was a 500 KB limit on the document size, while also having a very specific 2130 X 1346 resolution required for the image.

Why?

Then something occurred to me. I have yet to come across a government site that is not a form. This is true of every country where I have had to engage with a government website.

Every single government site is always a form.

In India, we regularly use government websites for Income tax, GST, Aadhaar, Driver’s License, Passport, and Voter’s ID. But even beyond this, birth/death certificate, stamp papers, duty payment and the list goes on.

Every single site is a glorified Form.

Someone figured out how to make a form online in 1992. Why are we still sucking at it?

None of these forms has any elaborate back-end computation. No AI. Not like you are uploading a picture and AI figures out the rest.

It is just a form which often does not even validate for data type. You can type text in the phone number field.

In India, all of the government sites are either designed by TCS or Infosys. Why can’t our IT behemoths deliver decent forms??

In the game of cricket, if a bowler were to take three wickets on three consecutive balls, or if a player were to score three goals in a football match, they were presented with a hat by the fans through a collection. It was a mark of respect offered to the individual who performed the feat.

Therefore, this came to be known as a Hat-trick.

The first known instance of this in cricket was in 1858, when H.H. Stevenson, who took three wickets on three consecutive balls and the fans collected money to buy him a hat.