I like LLMs. I really do. But my experience with them is very different from the Chicken Little folks.
Let's park coding to the side for a bit.
Case 1:
I am collaborating with a friend to build a graded Sanskrit reader for beginners using the Aesop's fables.
As a precursor, I asked Gemini 2.5 Pro if it had access to all the stories. Yes, it said. The three popular PD ones? Yes.
I asked it to print all three versions of a particular one, and it did. One of them was not the version it confidently claimed it was. We argued about it for a while. It shut up when I provided actual evidence.
I then decided to upload the three Gutenberg text files and asked it to use them as the source of truth to give me a list of unique stories putting variant plots, variant titles etc under the main heading. I gave it certain formatting requirements so that I could later verify if all 600-odd tales across the three books were properly accounted for.
Gemini tied itself into knots trying to do this. It could not guarantee that all the tales were present in the list it generated. It didn't know how to accomplish the task. Finally, I gave it a series of steps, an algorithm based on an n-branched tree. Only then did it manage to generate the list for me.
This took me four hours of wrangling across three different sessions.
Case 2:
I have been buying TASCHEN editions of impressionists and other classical artists. I wanted Gemini to compare various editions, give me the pros and cons so that I could pick a good edition to buy. By the time we came to Michelangelo it went nuts, hallucinating editions, ISBN numbers, page counts, authoritative urls, worldcat searches ...
This took about two hours.
There are more such amusing anecdotes. Some from DeepSeek as well.
I have tried LLMs with python, and typst and a few other things. Sometimes they work, sometimes they don't. They definitely do not write code the way I want them to. They will use OOP even if I specifically warn them not to.
LLMs are VERY good at translation and languages. I will give them that. But reasoning? I am not convinced. I will believe that LLMs are good enough to replace programmers when the Amodei siblings can operate their company only using LLM developers.
> "Too many calories" is a simpler problem to solve: increase physical activity, take less sugar, or take sugar less frequently.
From Taubes' Good Calories, Bad Calories:
> “To attribute obesity to ‘overeating,’” as the Harvard nutritionist Jean Mayer suggested back in 1968, “is as meaningful as to account for alcoholism by ascribing it to ‘overdrinking.’”
Obesity etc are hormonal issues. It is not simple calorie-in, calorie-out math. I have been on a diet, off-and-on, for 15 years, and I can notice craving for carbs increase after "slipping/cheating" by consuming sugary food. While I generally manage to avoid reaching for snacks, I can only imagine how people can struggle with this, more so when others around them are stuffing themselves with food rich in sugar.
For me, it's calories-in calories-out. I've been counting calories for two decades and learned a couple things about my particular case:
- If I consume more than 2,300 kcal per day, I gain measurable weight in weeks.
- If I consume less than 2,100 kcal per day, I lose measurable weight in weeks.
- Physical activity has little effect on my weight change.
- Age has little effect on my weight change.
- When losing weight, I must work out to retain strength. Moving around an extra 10 kg takes strength that is lost when the weight is lost.
In my case, any hormonal effects are secondary to the above. I've changed what I eat over the years and my weight didn't change if I consumed the same kcals. There are likely subtleties I am missing, but eating less works for me. YMMV
Sugar, carbs in general, is particularly problematic.
I know of diabetics who cannot give up on sugary treats with full knowledge of the consequences of their behavior, people who have seen others in the family lose eyes, limbs, kidneys and life to the disease. Not a single one of them will consume a block of cheese or butter if one were to set these in front of them. Or a plate of diced carrots or cucumbers. It is always the chocolates and chips and the cookies and the ice cream and the biscuits.
Over the last two decades, I have been 128 kg at my peak. I have also been 93 kg. I have noticed that carbs absolutely wreck my ability to maintain weight. You lose the will to say no to food.
I am in India. Festivals start in September and continue for the next few months. It is an unending caravan of carb-heavy food. I can very easily put on 10 kg in those months. 70K excess calories in 2-3 months is not a lot. This does not happen with high fat food, because you cannot eat those in large quantities. They have high satiety value.
> Physical activity has little effect on my weight change
The human body is insanely adaptive. The natural impulse, I believe, is to conserve energy. For different people, it will respond to continuous over- and under-eating as well as over- and under-exertion in different ways.
I sometimes binge on cheese and nuts or peanut butter when I'm not at my best, but even when I do this for weeks, and I'm talking an extra 2000 calories in a day which is easy with those things, I haven't gained weight (fat). I don't feel as good and I don't recommend it, but I think it's safer than eating sugar if one is going to binge. I had a couple years 10 years ago where I ate sweet things most days and did gain noticeable weight (fat). So it's not like I'm genetically not going to gain weight either.
As an aside, physical activity might not affect weight change as much but it will affect the fat/muscle ratio.
You are very lucky to be able to eat like that without weight gain!
The extra 200 calories would be more of the same, just bigger portions. I record everything by weight when eating at home, and have a good feel for calories per unit weight, but double check often. Cheeses are about 100 cals per oz, breads or snacks with little fat (pretzels) are 80-100 cals per oz, nuts are dangerously calorific at 160-180 cal per oz. Peanut butter is the same. I eat nearly everything, both good and bad for me, but keep track of calories and it's been working for me. I do eat ice cream occasionally, with full awareness of how much. My calorie limit is a target, but I enjoy life. The only thing I've eliminated completely is alcohol.
It means how much I eat has a much larger effect than physical activity. I can eat more calories in 5 minutes than I can work off in two hours of brisk walking.
> Can't say I ever ran across that in any serious context
It exists. Invaders do not necessarily invent things out of whole cloth. You widen existing fissures in society in order to control it. The gotra-jaati-varna system has a complex history but the caste system in place for the past century or so in Indian society and politics is a gift of the British. They were the ones who started recording it during the census. A 2011 census found that people claim to belong to 4,673,034 different castes/sub-castes/gotras/clans.[1] Which is absolute insanity.
Castes that are considered to be "upper" in one district/state move down the ladder in other districts/states. But all this complexity is thrown out of the window in ideological discussions.
Flat Earth, Young Earth, Lysenkoism, Vaccines causing Autism, Alien Lizard People wearing Human Skin, et al are all examples of other beliefs that exist, have had books devoted to them, have true believers, etc.
These are also things that I don't come across in serious contexts.
> Invaders do not necessarily invent things out of whole cloth. You widen existing fissures in society in order to control it.
Sure, something seen in history often enough. This supports the notion of a caste system existing prior to the British.
> They were the ones who started recording it during the census.
If it didn't exist prior to the British "recording it during census" then what was it they were recording?
Something they made up whole that never existed prior?
> A 2011 census found that people claim to belong to 4,673,034 different castes/sub-castes/gotras/clans.
.. and this was the gift of the British? Sounds more like a home grown Indian expansion.
Not like flat earth. Something that would be discussed in academic contexts and the intellectual mainstream.
Articles of the last decade or two tend to be weaker in their claim, salvaging parts of the stronger version from the mid-century. The debate has since subsided.
The British absolutely had a role. One theory is the West needed a pliable state on India's west to guard the Arabian Sea from the Russians. A lot of Hindu politicians of the era were socialists. Nehru, the first Prime Minister of India, was a Fabian Socialist. What if India allied with Russia? Pakistan soon became a CENTO treaty member.
The thousand year Hindu-Muslim rivalry is the main driver, however. During the 16-18th centuries, European expeditions started hitting Indian shores, initially nibbling around the edges, and Muslim power on the subcontinent was comprehensively demolished by the Maratha Empire. The British drove out other European claimants and managed to wrest control over the subcontinent piece-by-piece.
Muslim elites did not mind ruling over majority-Hindu provinces for centuries. But loss of power and membership of a numerical minority creates insecurities and the idea of a separate Muslim nation started post 1857.
People can quibble about the origin of the idea and who is to blame. But no one can deny the fact that repeated massive riots between these communities (which are separate communities/religions, not ethnicities) in the decades leading to the partition signaled the inevitable.
> People can quibble about the origin of the idea and who is to blame.
Look... I think there is a whole lot of extremely motivated reasoning. Even the British weren't that keen to defend British imperialism by the time of India's independence. The Brits being already responsible for so many sins, make an easy place to put "blame" that cannot be borne elsewhere.
The alternative is to "blame" nationalism, self determination, liberalism. republicanism, and/or democracy.
Everyone knew that India was going to civil war. It was a big factor delaying British decolonisation. Gandhi's whole ideology/rhetoric as independence loomed changed to try an prevent it. Naive in retrospect.
This has always been the sticking point, for modern nationalism. People have strong, inconvenient notions of who the "nation" is.
The nasal "m" takes on the form of the nasal in the row/class of the letter that follows it. As "ñ" is the nasal of the "c" class, the "m" becomes "ñ"
Writing Sanskrit terms using the roman script without using something like IAST/ISO-15919 is a pain in the neck. They are going to be mispronounced one way or the other. I try to get the ISO-15919 form and strip away everything that is not a-z.
So, सञ्चिका (sañcikā) = sancika
You probably want to keep the "ch," as the average English speaker is not going to remember that the "c" is the "ch" of "cheese" and not "see."
Not unless it precedes a classless letter or it is actually "m."
All nasals becoming anusvaras is something Hindi/Marathi and other languages using the Devanagari script do. Sanskrit uses the specific form of the nasal when available.
> Switzerland has 9 million total people, and more millionaires than India and Pakistan combined.
Well, based on official figures, perhaps. I am pretty sure, however, that India alone has more than 20 million dollar millionaires.
A large part of the Indian economy operates outside the tax net. Some of it is deliberate (income from farming is not taxed, even if you earn tens of millions of rupees from your acreage), the rest is due to garden variety tax fraud.
Basic apartments in the top 10 cities start at $200,000 (could be something like $500,000 in places like Mumbai). And plenty of people own 2-3-4 of these. You have to move to distant suburbs if you want cheaper housing.
Millions of Indian households hold large quantities of gold. Millions invest in the stock market.
An upper-middle-class life for a family of three in one of the major cities requires around $30,000 easy (double that if you are the splurging kind) which can buy you a LOT more than what the same sum can get you in LA/NY/London.
Once you add these up, my figure is completely within the realm of possibility.
20 million dollar millionaires means that more than 1% of Indians have 1m USD, which is certainly not the case - the number of Indians actually paying income tax alone is just around that number.
India actually has about 320k USD millionaires, Pakistan 20k and Switzerland 370k, from a cursory Google search.
You forget the fact that most Indians don't live in the city in 1-2 Cr (roughly $120-250k) homes, and their gold jewelry usually amounts to $50k tops.
You only need about 5-10% of them to do so to start hitting the numbers from my claim.
> the number of Indians actually paying income tax alone is just around that number
The number of businessmen who avoid taxes is a significant minority.
> India actually has about 320k USD millionaires
Based on the number of people I know who own multiple properties (which makes them dollar millionaires even if you ignore everything else they own), I find it VERY difficult to believe this figure.
If 8% of the US population can be millionaires, I don't see why 1.5% of the Indian population cannot.
We may not agree upon a number, but I believe the figure is much, much larger than what the official figures state.
> You only need about 5-10% of them to do so to start hitting the numbers from my claim.
Dude, a bit more than 1% of Indians pay ANY form of income tax. Assuming that the number of corrupt politicians and people laundering black money is another 15 million (a very very far stretch), that's still 2% of the entire population.
> The number of businessmen who avoid taxes is a significant minority.
Thanks for proving my point.
> Based on the number of people I know who own multiple properties (which makes them dollar millionaires even if you ignore everything else they own), I find it VERY difficult to believe this figure.
Your anecdote means shit compared to statistical data that says otherwise.
> If 8% of the US population can be millionaires, I don't see why 1.5% of the Indian population cannot.
Because again, India is a highly unequal country, with very few opportunity for the brightest and most entrepreneurial. Why has India not manufactured even one hard tech startup yet? Why is it so hard to get even one manufacturing unit up without the blessings of some babus and Lalas? That's why India is poor and unequal. No surprise why people who move out of India tend to be more successful, especially those who migrate to the US.
> We may not agree upon a number, but I believe the figure is much, much larger than what the official figures state
By that respect, Switzerland would still have a lot more millionaires simply by virtue of the number of people parking their assets there. There are no official figures for that either.
UBS estimated that there were 0.86M dollar millionaires in India at the end of 2023. Credit Suisse, in a 2022 report, estimated that the number would double from 0.79M in 2021 to 1.6M in 2026. Different reports use different standards. A lot of them exclude the primary residence in their computations because (maybe) they are looking for people who have free assets that these wealth managers can manage on their behalf.
In my claims, I do not. An asset is an asset.
> Dude, a bit more than 1% of Indians pay ANY form of income tax.
Wealth and income are not the same thing, though there may be some correlation. A person can have large parcels of land and still have income below the IT threshold ($30,000 pa for a couple according to the latest amendments).
Why do you think the IT department tracks random things like electricity bill payments over Rs. 100,000 pa? They want to know about people who have absurd spending power but who do not show corresponding income in their returns.
> without the blessings of some babus and Lalas
The bureaucracy and red tape is a problem and a huge drag on the GDP. We need serious bureaucratic reform. It is a wonder that so much wealth exists in spite of these problems.
I am not sure that the intersectional feminist writing this essay has access to the mental framework necessary to understand the fundamental dispute between the countries. But her toxic essay needs some grounding. So...
Pakistan was created upon the idea that Hindus and Muslims are two distinct nations that can never live together in peace. Unfortunately, total population exchange was never achieved due to Indian politicians of the era living in la-la-land (a majority of them continue to do so to this day). Nehru, for all his claims to erudition, was supposedly horrified when Hindu-Muslim riots restarted post-partition.
Everyone knows why these disturbances happen across space and time. It is not a question of a parcel of land ("Kashmir"). It is a question of jihad. The Pakistani Army is a jihadi army. These are the words of General Zia-ul-Haq, dictator/president of Pakistan from the foreword to the book "The Quranic Concept of War" by Brigadier S. K. Malik:
> I write these few lines to commend Brigadier Malik’s book on ‘The Quranic Concept of War’ to both soldier and civilian alike. JEHAD FI-SABILILLAH is not the exclusive domain of the professional soldier, nor is it restricted to the application of military force alone.
> This book brings out with simplicity, clarity and precision the Quranic philosophy on the application of military force, within the context of the totality that is JEHAD. The professional soldier in a Muslim army, pursuing the goals of a Muslim state, CANNOT become ‘professional’ if in all his activities he does not take on ‘the colour of Allah.’ The non-military citizen of a Muslim state must, likewise, be aware of the kind of soldier that his country must produce and the ONLY pattern of war that his country’s armed forces may wage.
> I have read this book with great interest and believe that it has a useful contribution to make towards this understanding that we jointly seek as citizens of an Islamic State, soldier or civilian. I pray and trust that this book will be read by many. For a task so sincerely undertaken and so devotedly executed, the author’s reward is with his Lord.
For some Pakistanis, Muhammad bin Qasim is the first Pakistani because he destroyed the Hindu kingdom of Sindh in the early 700s. For others, Pakistan is the new Medina, a place to gather one's strength and attack Mecca ("India") when ready, thus following in the footsteps of their prophet. Some Pakistanis are unwilling to accept that their ancestry is Indian. Instead they go looking for their presumptive fathers in Persia, Arabia and, more recently, Turkey.
For decades, Pakistanis have named their missiles and armaments after Islamic invaders from the north-west who raped and murdered their ancestors, forcibly converted them to Islam and sold their women into sex slavery. The list of their jihadi antics is long.
Just as there can be no peace between cannibals and humans, there can never be peace between the two countries because jihad does not allow for peace: the enemy must be conquered and converted or destroyed. And India cannot allow that.
I read it to see if the evidence shown at the two press conferences by the Indian DGs of Operations and Modi landing on the Adampur Airbase (which was supposed to be heavily damaged) and giving a speech with the S-400 system (which was supposed to be destroyed) behind him was covered in the article.
It isn't.
The only thing covered is claims by a state known for its pathological lying and its all-weather ally. So, more of a press release than journalism.
This is fine. I was referring to actual evidence. The Pakistani generals were blatantly lying in their PCs. All the fact checkers in the world suddenly disappeared. And the evidence presented on the Indian side which can be cross-checked using satellite imagery is ignored.
And they have the audacity to "both sides" the event.
The question that needs to be answered is: if Chinese weapons are so great, how was India able to hit deep within Pakistani territory AT WILL? Or, is it the claim that this never happened?
> there's no evidence can prove Rafales have been shot down
Even if a couple of planes are lost, so what? You lose lives and equipment when fighting a war.
The Americans lost three F/A-18 Super Hornets in their expedition against the Houtis (who do not even have an air force). They are struggling to hit Houthi air defenses with their top-of-the line aircraft and almost escaped being hit multiple times.[1] I don't see the Americans whining about it.
India has the money to replace any losses and buy more from any supplier in the world (we also have a homegrown defense industry). The Pakistanis are stuck with the export version duds that the Chinese supply them.
I agree, india can still win even if a couple of planes are lost
in fact i think it's a very good mindset as it helps avoid further escalation of the war
> India has the money to replace any losses and buy more from any supplier in the world (we also have a homegrown defense industry). The Pakistanis are stuck with the export version duds that the Chinese supply them.
Agree, India has the money to buy 6 more Rafales, however, I believe the J10CE might offer better cost-performance and could be worth considering
Probably a mistake on India's part, but the future will tell us.
We have the habit of squandering away the gains made on the battlefield on the negotiation table, famously after the 1971 war that birthed Bangladesh where thousands of square kilometers of captured territory and 93,000 POWs were given up without forcing the Pakistanis to hand over POJK and GB.
This is without considering historical examples when Islamic raiders would lose battle after battle and Hindu kings would let them go back till the day the Hindus lose and are slaughtered enmasse.
Some good things have come out of this:
* We now know that Chinese radars do not work against Indian supersonic/hypersonic missiles.
* The Indian air defense systems proved themselves by shooting down drones, missiles and aircraft.
* India has changed its soft policy on terror by considering an terror attack on Indian soil to be an act of war.
* India is going to starve them of water. The suspension of the IWT stands.
This is a temporary pause. It is Pakistan's psyche to go running to America for help each time it has been spanked by India. And once it recovers, it tries another "misadventure."
While it is in India's interest to dismember Pakistan into four parts, it is in America's interest to keep the irritant alive. The US let China grow into the behemoth that it is. And it does not want to see this being repeated again with India.
In a way, yes. Hinduism slowly lost its warrior ethos post Buddhism and Jainism and their fetishization of non-violence.
All our books (itihasas and shastras) warn about enemies and how to deal with them. And people simply forgot all that.
As far as the opponents go, the religion doesn't really matter. Kings spared Mihirakula[1] of the Huns (who was a Shaivite Hindu) just like they spared various Islamic/Turkic raiders. And the results were as expected: betrayal and murder.
OT. I was watching Radhavallabh Tripathi's 3-episode series on Kālidāsa's Mālavikāgnimitram where he covers the politics of that era. And this point comes up. You can find it on YT (in Hindi) if you are interested.
* The History and Culture of the Indian People - is a 11 volume encyclopedia edited by R.C. Majumdar. Slightly dated because of when it was published, but it is a scholarly work that will give you plenty of references to other material if you want to dig deep.
* Sri Aurobindo & Swami Vivekananda - Their collected works are available online and you can pick and choose what you want to read because their output was vast.
* Books by K S Lal - historian who wrote a lot of books on the Islamic/medieval period
* Books by Meenakshi Jain - historian
* Books by śatāvadhānī R Ganesh - his Kṣāttra - The tradition of Valor in India is quite good
* Books by Sandeep Balakrishna on the Delhi Sultanate period, Tipu Sultan etc
* Books by Pavan K. Varma
* The Ramakrishna Mission publishes many books on Hinduism. "The Essentials of Hinduism" by Swami Bhaskarananda is a concise introduction to the religion.
There are hundreds of such books. Will add another comment if I can think of them.
The best option is to read good translations (can be difficult to find ones without bias) of the upanishads, itihasas and puranas. Everything else is based on them.
I'm not Indian, nor Pakistani, nor from South Asia.
I just wanted to clarify some points...
* It was Pakistan that shot 3 to 5 Indian jets. Pakistan says three Rafale jets, a MiG-29, and an Su-30, along with one Heron drone. NY Times confirms at least 3. This was a huge win for Pakistan by all military expert accounts. This was all using Chinese technology.
* Trump said today that it was India that went to the US to solve the situation. CNN also reported this.
To me, it was a huge win for China and Pakistan. A US official has also confirmed that, stating they stood by to see what China had.
India isn't innocent either; they meddle a lot with Balochistani separatist groups in Pakistan. This is no hidden secret. These terrorists have killed Chinese engineers.
There has been a lot of false information out there, particularly from Indian media (many accounts said a pilot was also captured, and other cities such as Karachi were hit). It was very alarming to see such blatantly wrong accounts by mainstream Indian media.
> was Pakistan that shot 3 to 5 Indian jets. Pakistan says three Rafale jets, a MiG-29, and an Su-30, along with one Heron drone. NY Times confirms at least 3. This was a huge win for Pakistan by all military expert accounts. This was all using Chinese technology.
Can not comment anything on this as it is not confirmed or denied from Indian side.It is only said that no pilot has been lost.
> Trump said today that it was India that went to the US to solve the situation. CNN also reported this.
You can see Trump's change of wording on the recent tweets as well as India clearly denying role of Trump as mediator.India said the discussions took place directly between the two neighbours-in-conflict.
> India isn't innocent either; they meddle a lot with Balochistani separatist groups in Pakistan. This is no hidden secret. These terrorists have killed Chinese engineers.
Is there any proof for this?I have not seen any.From what I have heard, they are supported by Afgans.Still no proof.
> There has been a lot of false information out there, particularly from Indian media (many accounts said a pilot was also captured, and other cities such as Karachi were hit).It was very alarming to see such blatantly wrong accounts by mainstream Indian media.
Yes, the examples given are fake news,peddled by Indian media.They do not care anything besides high TRP.
Let's park coding to the side for a bit.
Case 1:
I am collaborating with a friend to build a graded Sanskrit reader for beginners using the Aesop's fables.
As a precursor, I asked Gemini 2.5 Pro if it had access to all the stories. Yes, it said. The three popular PD ones? Yes.
I asked it to print all three versions of a particular one, and it did. One of them was not the version it confidently claimed it was. We argued about it for a while. It shut up when I provided actual evidence.
I then decided to upload the three Gutenberg text files and asked it to use them as the source of truth to give me a list of unique stories putting variant plots, variant titles etc under the main heading. I gave it certain formatting requirements so that I could later verify if all 600-odd tales across the three books were properly accounted for.
Gemini tied itself into knots trying to do this. It could not guarantee that all the tales were present in the list it generated. It didn't know how to accomplish the task. Finally, I gave it a series of steps, an algorithm based on an n-branched tree. Only then did it manage to generate the list for me.
This took me four hours of wrangling across three different sessions.
Case 2:
I have been buying TASCHEN editions of impressionists and other classical artists. I wanted Gemini to compare various editions, give me the pros and cons so that I could pick a good edition to buy. By the time we came to Michelangelo it went nuts, hallucinating editions, ISBN numbers, page counts, authoritative urls, worldcat searches ...
This took about two hours.
There are more such amusing anecdotes. Some from DeepSeek as well.
I have tried LLMs with python, and typst and a few other things. Sometimes they work, sometimes they don't. They definitely do not write code the way I want them to. They will use OOP even if I specifically warn them not to.
LLMs are VERY good at translation and languages. I will give them that. But reasoning? I am not convinced. I will believe that LLMs are good enough to replace programmers when the Amodei siblings can operate their company only using LLM developers.