Pritish Chakraborty @sub_modular

You're not ready for the big-O notation on this one: O(n+m+m*log(2+m/(n+1))+m^⅓*(n*log(n+2))^⅔)

Vals AI @ValsAI

2 weeks ago

We asked ten Claude Opus 5.5 agents to devise a faster shortest-path algorithm and prove it in Lean. Within 15 hours, they produced C-HD: a formally verified improvement over the published bounds.

75 247 3K 1.4M 1K

@swaroopnath6 @Support @X need some help here. I can vouch for the fact that this account is hacked and not being run by the actual person.

Posting with friend's permission: @swaroopnath6's account has been hacked recently. Please ignore all messages and posts.

🚨 Live update: Karjat records a mammoth 112 mm of rainfall in just ONE HOUR ⚡⛈️ An intense cloudburst has been reported over the region #MumbaiRains

I must be among an extremely small group of people (n=1?) that have both 1) trained a frontier LLM and 2) designed and synthesized custom viruses in a lab with my own two hands. And I think that the takes on AI killing us all by creating dangerous viruses is total bogus.

A lot of people are missing Terence Tao’s point and thinking “mathematicians are upset that AI is better than them.” That’s not what he’s saying, and some people are forgetting that Tao is one of the most AI-pilled mathematicians out there. His point is that when people work on discovering something, along the way they invent new concepts. Those concepts later become useful far beyond the original goal, and enables further inventions. Finding a solution does matter, but the intermediate idea is often what makes the field richer, because other people can share it and build the next thing from it. In tech, we can use the analogy of collaborative software. We started with algorithms for merging changes in a Word document, and evolved that to concepts about versions, diffs, and merges, and later to real-time collaboration tools like Git, Google Docs, and Figma. Humans built upon these concepts and developed more powerful solutions. Terence’s worry is that a machine automating a solution robs the field of the value of developing the intermediate discoveries in the pursuit of larger discoveries. When automating a solution, the intermediate discoveries and invention of concepts can be buried or completely hidden in the black box. We don’t learn from them to build the next thing; it’s like we never made the invention of collaborative document editing and thus could not have the conceptual understanding to invent the next version – and since it’s hidden, we also don’t socialize them to allow other people to invent, too, a core tenet of collective discovery. So then, in both code and math, this leads to the atrophy of development of concepts in the field. In other words: pure ‘solution extraction’ that hides the process of discovery can leave the field with a checked-off theorem but little new insight or new questions to pursue. And it might prevent us from understanding a field deeper. I am seeing, first-hand, that atrophying of skills in software development. We push buttons and get solutions. There is much less incentive to develop new concepts and human skill. The bet most software companies are making is that LLMs are so effective in writing code that you’re still shipping overwhelmingly more value even with human skill atrophy, and it’s the right bet IMO. However, much of the software industry is built upon building things, not necessarily novel invention and research. In such an environment, you can say that you accept some atrophying of conceptual invention and human skill for more output. On the other hand, sectors like math and pure sciences that are focused on invention and insight might be the hardest hit by this. Practical/applied sciences might fall somewhere in the middle. An Alzheimer’s cure, room-temperature semiconductor, or highly effective carbon capture solution are far too valuable to sandbag and say only humans can do that to develop concepts in the ‘proper’ way. The outcome matters too much to treat the preservation of concept invention as the highest goal. Even there, though, hidden intermediates can slow the next breakthrough if nobody can see how the first one actually worked. So the question is not “is AI allowed to solve hard problems?” It is “in this field (math, science, tech, etc.), is the answer itself the main point, or are the concepts and abstractions we use to get there also the thing we need to maintain?”​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​​ In pure math, there’s an argument that the intermediates are often more useful than the solution, and atrophy in concept development is highly detrimental to the field. Solving Navier–Stokes, contrary to what some people claim, has little practical application, and pure math might be one of those fields where just finding a solution isn’t the entire point, and can actually be contrary to the field, which is what Tao is worried about.

spor @sporadica

3 weeks ago

as predicted, Terence Tao and friends are not happy Mathematicians that spend their lives trying to solve math problems are unhappy that AI is solving math problems because how dare they!

255 152 2K 899K 435

I was an intern under Seb in 2020. Unfortunately, the allegations of unscrupulous behavior is 100% believable. I am glad that the mask is off publicly. I really hope he doesn't weasel out of this. cims.nyu.edu/~tristanb/stat…

Sebastien Bubeck @SebastienBubeck

4 weeks ago

A series of false and inflammatory allegations against me are currently circulating on social channels. To clarify, I came into the discussion following academic norms, and I'm disappointed that it has come to this. Anyone who knows me knows that academic standards are of the

349 150 3K 2.5M 840

Please consider mailing me (or any on committee) if you have applied but are not from mainstream college or discipline. No guarantees still but a second chance from potential algorithmic marginalization among 1000s. vihari[no space]piratla[at]gmail

Vihari @p_vihari

a month ago

Calling on all final year bachelors, masters and PhD students to Google DeepMind APAC Research Symposium, Oct 29-30, Bengaluru. You can learn about recent research, explore career opportunities. 🗓️ Deadline: August 31st, 2026 📷 Event Website: rsvp.withgoogle.com/events/gdmrs26/

1 4 37 5K 16

a model can never learn research taste because research taste is driven by spite

On a December morning in a Haryana village, a cooking gas cylinder caught fire in a house with eleven people trapped inside. A twenty year old neighbour picked up a ladder and went in. He got all eleven out. Women, children, one by one. And in the seconds after the last person was safe, the house exploded and killed him. Thirteen years later, his father is still fighting the government to admit it happened. His name was Saurabh Garg. It was the eighth of December, 2012, at Pillukhera in Jind district. The house belonged to his neighbours. He heard the screaming, woke up, and went out with a ladder. Understand the arithmetic of what he did. Every trip up that ladder was a decision to go back. He made it eleven times. He could have stopped at four, or seven, and no one would ever have called him a coward. He stopped only when the house was empty. Then the ceiling came down. His father, Chander Bhan, a trader, has spent the years since going from office to office. The District Collector recommended his son for a national bravery award within a week of the funeral. Letters went from the Prime Minister's Office to the Home Ministry. His name was read out in the state assembly. And then nothing. For thirteen years, nothing. The file was written about again in 2015. And 2022. And twice more in 2023 and 2024. Every time, it moved somewhere and stopped. His father has said out loud what he believes the reason is. That his son was not the child of a politician or a bureaucrat, and that his father is not a rich man. In 2023 he took it to the state human rights commission. Its finding, when it came, was blunt. It said the denial of recognition was not because of any lack of merit in the boy's bravery, but solely because of the dereliction and administrative apathy of the authorities. There is a statue of Saurabh Garg in his village. His neighbours built it. The country he died in front of has still not signed the paperwork. He saved eleven people in about the time it takes to read this. India has taken thirteen years to fail to say thank you.

One of our NeurIPS submissions unfortunately received a "vibe reviewer": not knowledgeable enough to articulate meaningful concerns, but confident in their sense of taste to arbitrarily declare "not convinced"

🇮🇳 India Tops the World at the 56th International Physics Olympiad 2026! 🥇🥇🥇🥇🥇India's young physicists deliver an extraordinary performance at #IPhO2026 in Colombia. All five members of the Indian team win Gold Medals. 1/3 @PMOIndia @DrJitendraSingh @HBCSE_TIFR @TIFRScience

This statement came out yesterday on Passport Seva Divas, and the internet has been burning since. Let me walk through what happened. Yesterday was the 14th Passport Seva Divas. June 24 marks the anniversary of the Passports Act of 1967, so every year the Ministry of External Affairs holds a press event to talk about passport services. This year they had a lot to celebrate. India issued 1.39 crore passports in 2025. They have 545 Passport Kendras across the country, up from 77 a decade ago. They launched chip-based e-passports with biometric data stored on a microchip. Dr. S. Jaishankar put out a warm message talking about Surakshit Passport, Sugam Seva, Sashakt Nagrik. Good news all around. And then an MEA official said this: a passport is a travel document, not a document of citizenship. Now here is the thing. Legally, the official is correct. The Passports Act of 1967 defines a passport as a document issued by the government to enable its citizens to travel internationally. Citizenship is governed by a completely different law, the Citizenship Act of 1955. The passport has always technically been a travel document in the eyes of the law. But practically for most Indians, the passport is the single most trusted piece of identity they carry. It has their name, their photo, their date of birth, and the "Republic of India" on the cover. When they go abroad, immigration officers at foreign airports look at it and decide whether to let them in or not. Every country in the world treats an Indian passport as proof that you are an Indian citizen. So when the MEA says it is not proof of citizenship, people have a very simple and very valid question. If not the passport, then what? The government has no answer to this question. Aadhaar is held by over 135 crore Indians. It is built into almost every government service. You need it for LPG subsidies, bank accounts, mobile connections, income tax returns, school admissions. But Section 9 of the Aadhaar Act of 2016 explicitly says Aadhaar is not proof of citizenship or domicile. The Supreme Court confirmed this again as recently as 2025 in a case involving Bihar's electoral rolls. The Election Commission went to court and said Aadhaar cannot be treated as conclusive proof of citizenship. The Supreme Court agreed. Voter ID is issued by the Election Commission of India. It is the document you use to vote in a democracy. Surely that proves you are a citizen? No. The Bombay High Court has ruled that Voter ID is not proof of citizenship. Foreign nationals have obtained Voter IDs through fraudulent enrolment. The document proves you are on the electoral rolls, not that you are legally a citizen. PAN card? Even foreign nationals on valid visas can get a PAN card. It is a tax identification document, not a citizenship document. Driving licence? It is issued based on residency, not citizenship. Ration card is a welfare card for subsidized food. It proves nothing about nationality. The Bombay High Court has ruled that owning an Aadhaar card, PAN card, voter ID, or even a passport are not proof of Indian citizenship, which is determined solely under the Citizenship Act, 1955. So what actually proves you are Indian? The legally correct answer is a birth certificate showing you were born in India, or a citizenship certificate issued under the Citizenship Act, or documentary evidence of descent from Indian citizens. These are the documents that hold up in court. But the problem is that about 38 percent of children under 5 in India do not have birth certificates. For large sections of the country, especially in rural areas, these documents simply do not exist. Births were never registered. Land records are incomplete or disputed. School leaving certificates have name spelling variations across documents because nobody was paying attention to consistency 40 years ago. This is exactly what happened in Assam. In August 2019, the final NRC list was published and 19.06 lakh people were left out of 3.30 crore applicants. These people, many of whose families had lived in Assam for generations, found themselves in legal limbo with no clear path to prove citizenship. Cases included a 108-year-old woman whose name had been in the very first NRC of 1951 and who had voted in India's first election, but was still declared a doubtful voter. Documents were rejected for spelling mismatches between two government papers. Family members were split, with some included and others excluded based on the same underlying documentation. A Kargil war veteran was detained and had to go to court to prove he was Indian. The MEA did not make a controversial political claim. They stated a legal fact. But that legal fact has now reminded 140 crore people that the document they trust most does not actually do what they think it does. And it opened a question that the government, courts, and Parliament have been quietly avoiding for years. If the passport is not a document of citizenship, does the government give it to non-Indians as well? And wouldn't this announcement create doubts in the minds of other countries? The government verifies that before issuing one. But once issued, the government is now saying that same document cannot be used to prove what was verified to get it. The fact of the matter is that India does not have a clean, universally accessible answer to this question. The courts have carved out what does not count. Parliament has not legislated what definitively does. And millions of people, especially those without clean documentation histories, live with that uncertainty every day.

Sidhant Sibal @sidhant

3 months ago

Passport is a travel document, not document of citizenship, say MEA officials

2K 1K 16K 5.4M 929

TIL about a race condition bug in a radiation therapy machine called Therac-25 that caused the deaths of 3 patients. Therac was a radiation therapy machine for treating cancer patients. It had two modes: electron beam for shallower treatment, and high-energy x-ray/photon for deeper treatments. The older models of the machine had hardware interlocks as safety mechanisms to prevent any bad configuration from ever happening. Then came Therac-25 that removed many of the hardware interlocks in favor of implementing them in software. There was a race condition in which if an operator entered the wrong mode (e.g. x-ray) and then switched it quickly within 8 seconds, the mode would not be updated. As a result, the patient could end up receiving massive overdose of radiation. Between 1985-87, six such accidents were known to have occurred, leading to the death of at least 3 patients. Wikipedia notes this as a case study in software engineering and the danger of engineer overconfidence.

Despicable behavior so far from Modi Govt and TATAs in the accident investigation. People claiming Boeing cover up should know that Boeing hasn’t managed to cover up its past failures. No safety bulletin has gone out for 787s based on this accident. This is a cover up to protect Air India and TATAs.

Tehxi @yajnshri

4 months ago

It’s been a year since the tragic Ahmedabad Air India plane crash This UK-based couple lost their son in that heartbreaking disaster. On the first anniversary of his passing, they returned to the crash site to remember him. No one loves you like your parents do❤️🥺

33 385 3K 117K 110

Sir, they have blocked access to the very AI models you said we should share instead of build

Nandan Nilekani @NandanNilekani

5 months ago

India doesn't need to lead the world in building the most advanced AI models. But it must lead in ensuring benefits of AI are widely shared. @rvenk and I have an op-ed in The @EconomicTimes economictimes.indiatimes.com/opinion/et-com…

1K 166 851 1.6M 523

Way back in 2019, I reported extensively about the Globarena Intermediate marks fiasco in Telangana. In fact, I had run a major campaign on my media company Mojo TV. We were the first TV channel up pick the story too. 23 students died by suicide, I spoke with all their families and it was truly heartbreaking. I remember one particular case where a single mother who worked as a farm labour was very proud of her bright daughter. The entire village came to support her education, everyone pitched in with books, commute and all expenses. When the Globarena marks mess up happened, she cried the whole night. Her mother told her it is okay, and she can go for reevaluation or write the exam again, morning the mother wakes up to see her daughter hanging in the middle of the house. I had demanded blacklisting Globarena, arresting the officials. But I guess companies like Globarena are pests that never die, they just transform into something bigger and dangerous and come back to haunt. Globarena now has become Coempt Edu Teck that messed-up the CBSE evaluation. Globarena in 2019 messed up results of 9.74 lakh students in Telangana. Out of this total, approximately 3.8 lakh students failed. 23 suicides were reported. Back then too Globarena failed to provide the answer sheets. After all this, Globarena transforms into Coempt Edu Teck, messes up the future of 18.5 lakh CBSE students across the country. It is almost like they got rewarded for bad behaviour. The officials who were giving approvals for CBSE just 74 days prior knew the history of Coempt Edu Teck. They knew the company is incompetent, but still decided to go ahead and give them the tender. Yes, the Globarena/ Coempt Edu Teck guys are fabulous at landing tenders. The names and associations of Coempt Edu Teck board now seem to run pretty deep in the highest offices. Unless the rot is burnt right from the root level, this won’t stop. All board members, directors, CEO, investors, government employees that tweaked the system to ensure Coempt Edu Teck got the contract should be dragged to the courts & never should be allowed near any business again. But in this country that is so much in love with corruption, I doubt if this will ever change. Coempt Edu Teck may soon get the tender for all exams across the country. PS: To all the students, one exam is not the end. Kudos to your fight. Keep fighting, never ever give up. And if you are feeling low, speak with friends and family or reach out to any helplines. Roshni Helpline, Telangana +91 81420 20033

This is an unbelievable piece of work by Sarthak and something that requires amplification. Let me explain what he found, in simple terms. Sarthak is a Class 12 student from the 2025-26 batch, one of the 17 lakh students whose answer sheets went through CBSE's new On-Screen Marking system. He spent days reading through CBSE's evaluation tenders, scraped all 576 tenders CBSE has issued, and tracked how the rules changed across three versions of the same tender. The core finding is that the company that won the contract to scan and grade 17 lakh students' answer sheets is Coempt Eduteck. Coempt used to be called Globarena Technologies. Globarena was the company behind the 2019 Telangana intermediate exam disaster, where software failures led to 3.8 lakh students getting wrong or missing marks, and 23 students died by suicide. A government committee found systemic failure and negligence. Six months later, Globarena rebranded to Coempt Eduteck. So a company with that track record won a contract to handle 17 lakh CBSE students. Sarthak's investigation is about how the rules were rewritten to let that happen. The tender was issued three times. > First tender, February 2025. It existed, then disappeared from the public GeM portal. Sarthak scraped all 576 CBSE tenders and this one was missing from the archive entirely. > Second tender, May 2025. Four companies applied including TCS and Coempt. All four failed the technical evaluation. Cancelled. > Third tender, August 2025. Coempt won. Between the second and third tender, a series of rule changes happened, and every single one made it easier for Coempt to qualify. Here is what changed, one by one. 01. The old rules disqualified any company with a history of abandoning work, failing to complete contracts, or financial weakness. The new rules deleted this clause entirely. Coempt's Telangana history stopped being a barrier. 02. The old rules disqualified any company that was "blacklisted earlier." The new rules changed this to "currently blacklisted." Because Globarena rebranded after Telangana, removing the word "earlier" effectively erased their past. 03. The rules required Rs 50 crore average turnover over three years. Coempt's exact average came to Rs 50.86 crore. They cleared the bar by less than 1%. Earlier, a smaller company had asked CBSE to lower the bar to Rs 30 crore for fairer competition. CBSE refused. So the bar was kept high enough to block small players, but sat exactly low enough for Coempt to scrape through. 04. Software maturity is measured on the CMMI scale, 1 to 5. The old rules required Level 5. The new rules dropped it to Level 3. Coempt is a Level 3 company. 05. The cooling-off period for engaging retired CBSE officials was cut from two years to one. This makes it easier to use recently retired insiders to influence the process. 06. The old rules required experience with large projects of at least 5 lakh students each. The new rules removed the student count and counted cumulative answer-book volume across small projects instead. Coempt has many small fragmented university contracts. This helped Coempt and hurt TCS. 07. The old rules required bidders to own their own data centre and disaster recovery centre on Indian soil. The new rules allowed third-party MeitY-empanelled cloud hosting. Coempt runs on AWS and Azure. This helped Coempt and hurt TCS, which owns its own data centres. It also means student data is no longer on sovereign, Indian infrastructure. 08. The old rules required the bidder to own or control the complete source code of its software. The new rules deleted this. Coempt's platform runs on Microsoft's proprietary IIS, which they don't own. 09. A last-minute corrigendum, issued right before bid submission, removed CBSE's own power to blacklist the firm if its software failed catastrophically. So even a Telangana-scale failure couldn't get Coempt banned from future government tenders. 10. The penalty structure shifted from punishing mistakes to punishing delays. The old rules fined the vendor for wrong scanning, merged pages, and unscanned books. The new rules dropped those and instead levied Rs 50,000 per day for delays. This incentivises rushed scanning over accurate scanning. 11. The old rules had a hard accuracy threshold, error rate not to exceed 0.5%. The new rules removed this number entirely. 12. The old rules specified proper book and robotics scanners. The new rules just say "sufficient scanners." The definition was vague enough that, as Sarthak notes, the scanning could be done with a phone on a stand. 13. On the security side, the contract required a VAPT (vulnerability and penetration test) certified by CERT-In before go-live, and a restricted beta phase before launch. The system clearly wasn't restricted, because the other researcher, Nisarga, was able to access it and find vulnerabilities four days before go-live. So the mandatory security audit appears to have been bypassed. These are more than a dozen rule changes, all between the failed tender and the winning tender, all pushing in the same direction, all benefiting the one company with the worst track record in the field. The security holes Nisarga found last week now have an explanation. The system was built by a vendor that was specifically allowed to skip the security certification, the source code ownership, the data sovereignty, and the quality thresholds the original rules demanded. Following things need to happen immediately; 1. An immediate CAG audit of the tender process. 2. A parliamentary debate on the topic. 3. An independent investigation into > Why the first tender vanished? > Why the disqualification clauses were deleted? > Why the turnover bar was held exactly where it was? > Why the security level was dropped? > Why the blacklisting power was removed at the last moment? Sarthak, this is genuinely exceptional investigative work. Far better than most journalists with full resources ever manage. Take a bow. :)

Sarthak Sidhant @sidhant_sarthak

4 months ago

CBSE has systematically rewritten its rulebook to favor Coempt Eduteck. check out the blog.

550 7K 19K 2.4M 2K

This is a case study in how India’s entire vulnerability disclosure framework is structurally broken. Let me explain; The first broken piece is the legal trap. When Nisarga “hacked” CBSE, he technically violated Section 66 of the IT Act, which criminalises unauthorised access to a computer system. There is no carve-out for security research in Indian law. None. The fact that he reported responsibly to CERT-In gives him zero legal protection on paper. The only thing standing between him and prosecution right now is CBSE choosing not to file a complaint. Compare with the US. The DOJ formally announced in 2022 that “good faith security research” will not be prosecuted under the CFAA. The UK has similar protections. The Netherlands codifies it. Singapore has formal coordinated vulnerability disclosure policies. India has none of this. Every researcher who reports a flaw is literally betting that the affected department is nice enough not to retaliate. Because of this, the best Indian security researchers work for foreign bug bounty programs like HackerOne and Bugcrowd because the legal risk of reporting to Indian government systems is too high. They find Indian vulnerabilities and stay quiet. We have no idea how many critical flaws exist in our digital infrastructure right now that researchers have spotted but won’t report. The second broken piece is CERT-In itself. CERT-In is mandated as India’s nodal cybersecurity body but operates with three structural disadvantages most people don’t know about. It has no enforcement authority. It can issue advisories. It can recommend patches. It cannot force a government department to fix anything. CBSE could ignore CERT-In for the next 5 years and there’s no mechanism to compel action. Its budget is roughly Rs 700 crore annually. For comparison, the UK’s NCSC operates with Rs 5,800 crore. The US CISA operates with Rs 25,000+ crore. India’s cyber defence per citizen spend is one of the lowest among G20 countries. The 2022 CERT-In directive that required companies to report incidents within 6 hours and store logs for 180 days generated huge industry pushback because the timeline was unrealistic. The directive remains active but enforcement has been patchy. It’s the opposite problem from CBSE. Over-regulation of the private sector, under-regulation of the public sector. The third broken piece is the architectural mistake repeated across every system. CBSE OSM had a 17-year-old find authentication bypass. Aadhaar had multiple leaks in 2018, 2019, 2023. CoWIN had a Telegram bot in 2023 pulling personal data using phone numbers. The Income Tax e-filing portal had multiple authentication issues during its 2021 relaunch. The common thread is the same architectural mistake. Government tech is built via tender. Lowest bidder wins. Security is treated as a nice-to-have. Systems go live with 6-month delivery deadlines and 1-month security audits. Patches happen reactively. We are investing heavily in our digital infrastructure. DIGIPIN is rolling out across India. Account Aggregator framework is integrating banking, insurance, mutual funds, and credit data into a single financial data exchange layer. ONDC is integrating thousands of small businesses and consumers. Each of these systems is bigger than CBSE OSM, but our approach to this is tepid. Israel pays its top security researchers Rs 1-3 crore packages through Unit 8200 and post-military startups. The talent stays in country and builds Check Point, Cyberark, Wiz, Palo Alto Networks. India trains 1.5 lakh cybersecurity engineers annually and exports most of them. The US runs Hack the Pentagon and Hack the Army, where they invite ethical hackers to legally test the most sensitive systems and pay $5,000-$50,000 bounties India’s version is closed-loop and opaque. We have no equivalent for Defence, Income Tax, or any major department. 1/2

nisarga @ni5arga

4 months ago

I had hacked CBSE's OSM (On-Screen Marking Portal) in February and had reported the vulnerabilities to CERT-In, but they were unable to patch most of them. I've written a detailed blog post about it here: ni5arga.com/blog/posts/hac…

217 1K 5K 751K 1K