Y
32 posts
I had two seniors who were crazy cracked
one of them was single digit in everything and 2x gold medal in icho - recently found out she is an swe at vc firm in usa rn
another of them was a madlad, never studied bio, got interested in bio in class 11/12 - but he didn't give neet, he got all exam double digit rank, joined top IIT cs, gave neet next year air 2, and he was one of the leading helpers during covid times for vaccines all over rthe world
Richa Sharma@richa_lq
I’ve been dying to reconnect with her since this person deleted all the social media when she turned 17 to hyper focus and I lost touch ever since she moved to the states for her undergrad. I look upto her so damn much & kinda just wanna work with her again. Last I heard she was at Stanford. I might just have to find her 😭 directly.
English
there are a lot of benchmarks that suggest 5.6 sol is the best model in the world right now, but the most reliable way to tell is that elon is obsessed with me again
English
sam altman is probably the greatest ceo of our time. he practically has every major tech CEO investing hundreds of billions of dollars to compete with him, sometimes even colluding (see elon + zuck). I don't know if gates, zuck, or even elon had this level of competition when they were forming their companies. OpenAI still consistently churns out some of the best models and still emerges as the winner in almost every category. my only question is what did @paulg see during that 10 min YC interview
English
I gotta start using this account to get free shit, right?
this is why we play the game
English
Asked my uncle for a referral a few months ago
He's an IIT Roorkee grad from the 1970s, works at a senior level in BARC, and has connections everywhere.
He said, Times have changed. I've seen people from ordinary colleges do great things. Your first job might be humble, but where you are in 2–3 years is on you. Build your skills. You don't need my referral
Fair enough !!
English
@sabeer Bhadwa sitting in USA and divorced. Your partner who was in hotmail is doing better than you. Chutia here’s free advice - if you want to progress don’t criticise your birth country. This is then a curse on you. You can never succeed further because of your attitude
English
@agrawalmanindra @ni5arga initially after this years' jee adv results i was a little sad that I won't be able to get a decent branch in iit kanpur campus. But after seeing so many of this guy's tweets I think god was protecting me from stupidity.
English
@ni5arga The problem with Telegram channel is not sharing of leaked paper, there are many other ways of doing it, rather that it can be used to spread fake news of leak that appears genuine. It was done by someone during JEE Advanced. It causes unnessecary confusion.
English
> can't stop paper leaks
> ends up blocking telegram
blocking telegram totally isn't even possible, telegram is designed in such a way which easily allows people to use proxies and other methods of circumvention.
National Testing Agency@NTA_Exams
NTA STATEMENT REGARDING THE ACTION ON TELEGRAM PLATFORM IN INDIA 1. The National Testing Agency (NTA) welcomes the directions issued today in respect of the Telegram platform in India. The directions, issued on recommendations of NTA are calibrated and bounded in time: (a) a direction under Section 69 A of the Information Technology Act, 2000, restricting access to the Telegram platform in India for a defined and limited period ending 22 June 2026, covering the day of the NEET (UG) 2026 re-examination and its immediate aftermath; and (b) a direction requiring the platform to disable, in India, the message-editing feature in respect of messages already posted, for a defined period ending 30 June 2026, addressing the specific structural feature through which the platform has been used to fabricate after-the-event “paper leak” evidence in respect of national examinations. Both measures have been taken in the interest of public order, in response to the organised use of the platform by cheating rackets to defraud candidates appearing for the NEET (UG) 2026 re-examination scheduled on 21 June 2026. NTA expresses its gratitude to the Ministry of Electronics and Information Technology for this timely action, in the interest of students, which will go a long way in helping NTA to be able to conduct safe and secure examinations on 21st June 2026. 2. Throughout the period leading up to the present action, the Indian Cyber Crime Coordination Centre (I4C), under the Ministry of Home Affairs, has served as the principal nodal agency coordinating the operational response to the Telegram-based fraud and misinformation targeting NEET (UG) 2026 candidates. Acting on inputs received continuously from NTA, from State law-enforcement agencies including the police forces of Bihar, Gujarat and Rajasthan, and from its own continuous monitoring of public channels and platforms, I4C has secured the prompt take-down of a substantial number of Telegram channels, groups and bots whose names and content openly advertised their fraudulent and misleading purpose. This was done with active support of Ministry of Electronics and Information Technology. NTA places on record that the intelligence-sharing and coordinated take-down action led by the Ministry of Home Affairs, through I4C, and MeitY has been continuous, prompt and substantive, and remains the operational backbone of the response. This sustained inter-agency effort, well in advance of the present platform-level action, is the reason the harm caused by these rackets has been contained to the extent it has. 3. The directions issued today by MeitY have been made following references by NTA and the Department of Higher Education, Ministry of Education, drawing attention to the structural limits of channel-by-channel action and seeking graduated platform-level compliance. The directions are a measure of last resort, taken only after intermediate remedies, including the take-down action coordinated by I4C, had been pursued and had not produced, at the platform level, the response required to protect candidates in the run-up to the examination. The calibration of the directions - a narrow platform-access restriction confined to the examination window, together with a feature-specific compliance direction for the post-examination period - reflects an effort to address the public-order concern with the minimum restriction necessary. 4. Over the preceding weeks, channels operating openly on the platform under names that themselves advertised their purpose - “PAPER LEAKED NEET”, “Re-NEET 2026”, “Private Mafia”, “REE NEET MAFIAA” and similar formulations - demanded sums ranging from a few thousand to several lakhs of rupees from candidates and their families, in exchange for purported access to the re-examination paper. NTA has placed on the record, and reiterates, that there is no such paper available outside the secured examination chain. The promise of any such material is, in every instance, a fraud. 5. The direction requiring Telegram to disable its message-editing feature in India through 30 June 2026 addresses a separate but related concern. The feature, in its present form, permits a channel administrator to edit the content of a previously posted message - including the substitution of attached files such as PDFs - while the original send-time stamp is retained. This capability has been used, in respect of multiple recent examinations, to fabricate after-the-event “paper leak” artefacts: a channel administrator edits an older, innocuous message to insert the actual question paper after the examination has been conducted, and the resulting chat is then circulated as purported “evidence” that the paper was in circulation before the examination. The MeitY direction closes this avenue of fabrication for the post-examination window in which such artefacts have historically been deployed. 6. Independent action by State law-enforcement agencies has, over the same period, reinforced the scale and seriousness of the concern. The Bihar Police Economic Offences Unit issued a formal public advisory on 9 June 2026, warning candidates against fraudulent claims of pre-examination access to the paper circulated through Telegram and other platforms. The Ahmedabad City Cyber Crime Branch arrested members of an inter-State cyber-fraud gang found to be operating eight Telegram channels in furtherance of the same modus operandi, with documented transactions of approximately ₹1.5 crore routed through fraudulent bank accounts and approximately one thousand mobile numbers contacted in a single month. Investigations are in progress in multiple other States. 7. NTA acknowledges that the access restriction issued by MeitY affects lakhs of citizens who use the Telegram platform for legitimate personal, educational, professional and informational purposes, and sincerely regrets the inconvenience caused to them. The access restriction is, by its express terms, confined to the period ending 22 June 2026 - i.e. the day after the examination. The feature-specific direction in respect of the message-editing function, which remains in force through 30 June 2026, does not affect ordinary use of the platform for sending or receiving new messages. 8. The NEET (UG) 2026 re-examination will be conducted as scheduled on 21 June 2026. The security of the examination is unaffected by the action taken; it is, in fact, the very purpose of the action. Every candidate and parent is reassured of NTA’s commitment to conducting a fair, secure and credible examination. Candidates are urged to focus on their preparation, to disregard unverified content circulating on any platform, and to rely exclusively on the NTA website (neet.nta.nic.in) and verified NTA handles for all examination-related updates. 9. Any encounter with fraudulent solicitations - in person, by telephone, or through any online platform - should be reported immediately to the National Cyber-Crime Helpline at 1930, or through the National Cyber-Crime Reporting Portal at cybercrime.gov.in. NTA’s own helplines remain available at 011-40759000 / 011-69227700 and at neetug@nta.ac.in. 10. NTA places on record its sincere appreciation of the Ministry of Electronics and Information Technology for the timely and calibrated directions issued, of the Ministry of Home Affairs and the Indian Cyber Crime Coordination Centre for the sustained operational coordination that has made today’s action possible, of the Central Bureau of Investigation for its parallel inquiry into the underlying offences, and of the police forces of Bihar, Gujarat, Rajasthan and other States for their independent enforcement action - each of which has contributed, in its own measure, to protecting the integrity of one of the country’s most consequential examinations and the interests of its candidates.
English
@shaashvats30 @agrawalmanindra exactly? if the data on which chebychev's is applied is itself skewed due to cheating, how will it be effective in pointing out outliers?
English
@agrawalmanindra Would it be possible for number of outliers to exceed that number predicted by the inequality? I am not able to understand how such a situation can arise.
English
Consider a hypothetical:
A student's answer key says he scored 200 in the first exam of JEE advanced. According to average and std. dev of the first paper, Chebyshev's inequality says this is well within the maximum possible score of 375 in the first paper. Now, is this score of 200 possible or a software error?
Chebyshev inequality doesn't say anything significant here
Manindra Agrawal@agrawalmanindra
Claims like below show a lack of understanding of basic statistics. When there are 60,000 students, some will have large gaps in marks for two papers. Chebychev's inequality quantifies it. It has been the case in JEE Advanced since two papers were introduced. 😊
English
Y retweetledi
Your model expects 15 people
41 people exist at 80 diff in p1-p2
2 people expected but 9 people exist in 100 diff
What does this prove ji
Saket Choudhary@saketkc
@munna_bhai_bsms @agrawalmanindra Values are here:
English
Y retweetledi
I don't think this is Chebyshev's inequality
the latter *upper bounds* the deviation probability in terms of the variance; it's a concentration phenomenon
the phenomenon you're pointing to is the opposite, anti-concentration
it's saying (correctly) that in a random variable with a moderately large variance, we expect to see outliers in a large enough sample
Manindra Agrawal@agrawalmanindra
Claims like below show a lack of understanding of basic statistics. When there are 60,000 students, some will have large gaps in marks for two papers. Chebychev's inequality quantifies it. It has been the case in JEE Advanced since two papers were introduced. 😊
English
Y retweetledi
use this upper bound chebychev for Gaokao & between JEE Mains & Advanced (P1 & p2)
Distance of actual from upper bound should be same for all test globally and between JEE Main & JEE advanced
Manindra Agrawal@agrawalmanindra
Claims like below show a lack of understanding of basic statistics. When there are 60,000 students, some will have large gaps in marks for two papers. Chebychev's inequality quantifies it. It has been the case in JEE Advanced since two papers were introduced. 😊
English
@agrawalmanindra @Bayesprof you missed another extreme situation called cheating.
English
@Bayesprof Such a large variation can occur only in extreme situations, e. g., mental fog, health issue, arriving very late in one exam etc. Data from precious years also shows similar pattern.
English
I have not seen the data and have absolutely no reason to doubt the assessment of @agrawalmanindra. However, a 100-point swing, if true, is indeed striking. @agrawalmanindra would love to see an analysis from the last few years to see patterns. A 100-point swing does show the tremendous stress students go through while taking the test. Brutal.
Manindra Agrawal@agrawalmanindra
Claims like below show a lack of understanding of basic statistics. When there are 60,000 students, some will have large gaps in marks for two papers. Chebychev's inequality quantifies it. It has been the case in JEE Advanced since two papers were introduced. 😊
English
@agrawalmanindra @sainivansh1708 @sigoen_sidd kabhi bolte hain data is with iit r fir kehte hain ki analyzed??
English
@sainivansh1708 @sigoen_sidd We do analyze. So far, nothing unusual has been observed.
English
Claims like below show a lack of understanding of basic statistics. When there are 60,000 students, some will have large gaps in marks for two papers. Chebychev's inequality quantifies it. It has been the case in JEE Advanced since two papers were introduced. 😊
Ananya Chopra@Ananya1669
JEE Advanced 2026 — a student scored -3 in Paper 1 and 104 in Paper 2. Same exam. Same day. Same syllabus. Paper 2 was rated harder this year. 60,000 students. Dozens of 100+ mark swings between P1 and P2. P1 ends 12 PM. P2 starts 2:30 PM. That's a 2.5hr window. #JEEAdvanced2026
English
@agrawalmanindra @sainivansh1708 @sigoen_sidd why cant cameras be placed in opposite side of screen then? your arguments are very illogical
English
@sainivansh1708 @sigoen_sidd Cameras were not used because they can capture screen and can relay it outside. As part of security policy, they are not permitted inside exam hall.
x.com/i/status/20639… has better analysis than me on this.
Saket Choudhary@saketkc
@pravesh @agrawalmanindra Hi @pravesh @agrawalmanindra, here's a more sketched-out version of P1-P2 with the caveat that the extremes are included in the estimation. There are 2x more outliers above 80 and 4x more above 100 than would be expected by chance. Nothing survives once you do bonferroni.
English
@agrawalmanindra @SinaYanka security guard is sleeping due to security reasons
English
@SinaYanka CCTV recording is not done due to security reasons.
English
what is required is to look at cctv recordings, adjacent candidate scores and rule out malpractice.
even if it's a 1000 cases, it can be done in 3 days by a team of 10 people
Manindra Agrawal@agrawalmanindra
@PosterInternet The inequality is indeed in wrong direction but shows that occurance of ouiers is not surprising. What is needed is more refined argument that involve looking at the overall distribution.
English
@agrawalmanindra @Travidscot marks in both papers like 70 and 155 are believable...but -3 to 114? you seriously believe it? don't you think its suspiscious?
English
@Travidscot A large number of students with 100+ gap between marks in two papers for example. We do many other analytics too.
English
@agrawalmanindra sir but if we use conditional probability to calculate the marks of these students in 1 paper based on their marks on another paper wouldn't it be very close to 0?
English
@agrawalmanindra @kvvsgopalrao why are you trying to provide umbrella and shelter to cheaters when students are giving first hand experience of irregularities and cheating happening at their own centers?
English
@kvvsgopalrao Your start with fundamentally wrong assumption. The dtatset is not "ours". It comes from a site that estimates student scores.
English
IIT Directors and Statistics professors are failing 5th-grade math to cover up a massive cheating scandal in JEE Advanced 2026.
They claim 100-mark jumps between Paper 1 and Paper 2 are just "normal statistics." I ran their own data. They are lying.
A thread. 🧵👇
AI was used
Manindra Agrawal@agrawalmanindra
Claims like below show a lack of understanding of basic statistics. When there are 60,000 students, some will have large gaps in marks for two papers. Chebychev's inequality quantifies it. It has been the case in JEE Advanced since two papers were introduced. 😊
English