Why is Pakistan cleaning up PoJK now?
Why is it important? Connected events.
1. The moment US steps up in Tibet, there'd be HUGE backlash from India, Nepal & China. (Don't ask me why!)
2. Pahalgam went wrong due to disproportionate action on Kinara.
3. No trade deal with Trump regime yet.
4. Sergio is trying his best to divert GoI's attention.
5. US wants to pull India into its sphere of influence, when it comes to energy, so that it can "throttle" India's growth. Something they failed to do with China. Hence, Venezuela.
6. Marco was accorded coldest welcome. First sign of India's "NO" to US plans in PoJK.
7. China's sudden thaw with India, is calculated one.
8. Pakistan is doing cleanup operation for US to setup base in PoJK / GB with Tibet in mind.
9. HH Dalai Lama moved to Delhi and then, he has plans to go to Ladakh for extended vacation.
10. With some in #CTA #Tibet already getting closer to US, after new CTA last week, US thinks (for whatever reason!) it could enter Tibet via Nepal.
I've said this before. But, saying it again.
The moment US puts boots on ground in Kashmir, something catastrophic would happen, for US.
Either a sleeping elephant will wake up. Or, a dead man will start walking.
"India Has Labs, Money, And Startups — Here’s Why They Struggle To Work Together"
@mchellap & @astrokaran write for @SwarajyaMag — the structural diagnosis, the global models that work, and a specific proposal to fix India's R&D architecture.
https://t.co/E28ieMQBsi
As narrated.
I was at Pehalgam on 22 Apr 25, was lucky not to be at Besaram valley. I was in hotel room about 3 kms from the place of attack,saw the news on TV screen. My wife &I huddled as we remembered Mumbai & wondered if more such terror groups r searching tourists like us.
1/2 The architectural foundation of modern defence capability is not a platform, it's the stack. Like the sw stack that runs mobile phone or banking system, the defence stack separates sw from hardware so each can evolve independently.
It's a weird time. I am filled with wonder and also a profound sadness.
I spent a lot of time over the weekend writing code with Claude. And it was very clear that we will never ever write code by hand again. It doesn't make any sense to do so.
Something I was very good at is now free and abundant. I am happy...but disoriented.
At the same time, something I spent my early career building (social networks) was being created by lobster-agents. It's all a bit silly...but if you zoom out, it's kind of indistinguishable from humans on the larger internet.
So both the form and function of my early career are now produced by AI.
I am happy but also sad and confused.
If anything, this whole period is showing me what it is like to be human again.
We use AI coding tools across the company - with the caution to engineers that they have to review and approve all the code and take responsibility for it. We don't force feed or mandate AI tools, and leave the decision on how best to use AI to experienced engineers.
With that said, I spoke to one of our most productive engineers, with 2 decades of experience. He is doing some critical UI work for his product, very challenging to get right performance wise.
He told me that he is now shipping features that would have taken him 3 weeks of work that he got done in a day. AI needed him to provide the structure, it fleshed out the details and that required him to draw on his experience.
One of his less experienced team members has built an internal tool with AI that is being used across other teams. He said that team member could not have built the tool without AI.
Overall, for UI work, he is very bullish on AI.
All these are independent of my own research project. Just reporting on our collective experience.
I’m an Indian Muslim, and let’s settle this:
Vande Mataram was sung by Muslims in 1905, banned by the British for uniting us, quoted by Maulana Azad with pride and only became “un-Islamic” when Jinnah needed a communal rift to justify Pakistan in 1937.
So spare me the theatrics in 2025.
This isn’t about Islam.
It’s about politicians recycling a 1937 separatist script because victimhood still pays.
India is my homeland.
I don’t need London, Lahore or Lutyens liberals to tell me what patriotism looks like.
@narendramodi@grok please suggest what are the problems solved for these 100cr with Social security net? Funds moved from which scheme to which scheme? Does it benefit market and economy?
@svembu@rajuvegesna Sirji, Are you again the person who registered these app store accounts in US, and left them to this controversy?
Proud of Zoho! Proud of Arattai!! (Looks like a tounge twister for Non-Tamils)
@Yogakshema_ It's not about one Failed Marshal!!
I see Mahavir & Kirti Chakra equivalent in Pakistan "Hilal I Jurat" became "Sab ki Jaroorat hota hai". Except 2, all other awardees in History of Pakistan are General ranks!
Pakistan citizen, diaspora to take such failed nerves to control.
@coldhealing@grok can you summarise this thread and other comments? Which specialization matters in this current shifts, and why? You can go through my specialization and suggest something that's more tuned to my strengths. Leave definitive response than speculative.
We're missing (at least one) major paradigm for LLM learning. Not sure what to call it, possibly it has a name - system prompt learning?
Pretraining is for knowledge.
Finetuning (SL/RL) is for habitual behavior.
Both of these involve a change in parameters but a lot of human learning feels more like a change in system prompt. You encounter a problem, figure something out, then "remember" something in fairly explicit terms for the next time. E.g. "It seems when I encounter this and that kind of a problem, I should try this and that kind of an approach/solution". It feels more like taking notes for yourself, i.e. something like the "Memory" feature but not to store per-user random facts, but general/global problem solving knowledge and strategies. LLMs are quite literally like the guy in Memento, except we haven't given them their scratchpad yet. Note that this paradigm is also significantly more powerful and data efficient because a knowledge-guided "review" stage is a significantly higher dimensional feedback channel than a reward scaler.
I was prompted to jot down this shower of thoughts after reading through Claude's system prompt, which currently seems to be around 17,000 words, specifying not just basic behavior style/preferences (e.g. refuse various requests related to song lyrics) but also a large amount of general problem solving strategies, e.g.:
"If Claude is asked to count words, letters, and characters, it thinks step by step before answering the person. It explicitly counts the words, letters, or characters by assigning a number to each. It only answers the person once it has performed this explicit counting step."
This is to help Claude solve 'r' in strawberry etc. Imo this is not the kind of problem solving knowledge that should be baked into weights via Reinforcement Learning, or least not immediately/exclusively. And it certainly shouldn't come from human engineers writing system prompts by hand. It should come from System Prompt learning, which resembles RL in the setup, with the exception of the learning algorithm (edits vs gradient descent). A large section of the LLM system prompt could be written via system prompt learning, it would look a bit like the LLM writing a book for itself on how to solve problems. If this works it would be a new/powerful learning paradigm. With a lot of details left to figure out (how do the edits work? can/should you learn the edit system? how do you gradually move knowledge from the explicit system text to habitual weights, as humans seem to do? etc.).
@grok@NikkeiAsia@grok What are the technological gaps you see from India side on EV Manufacturing? How is India handicapped compared to other SEA countries?
What are examples from other countries that Indian Govt can take in becoming the best place for EV Manufacturing? How do we incentivize?