<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[Stories by Prayag Mohanty on Medium]]></title>
        <description><![CDATA[Stories by Prayag Mohanty on Medium]]></description>
        <link>https://medium.com/@topper1728?source=rss-78ce477380be------2</link>
        <image>
            <url>https://cdn-images-1.medium.com/fit/c/150/150/1*P1OB8FO9DXzECu2ZLCdZIQ.jpeg</url>
            <title>Stories by Prayag Mohanty on Medium</title>
            <link>https://medium.com/@topper1728?source=rss-78ce477380be------2</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 23:01:14 GMT</lastBuildDate>
        <atom:link href="https://medium.com/@topper1728/feed" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[Cold Caves, Hot Waves]]></title>
            <link>https://medium.com/@topper1728/cold-caves-hot-waves-a012107960b9?source=rss-78ce477380be------2</link>
            <guid isPermaLink="false">https://medium.com/p/a012107960b9</guid>
            <category><![CDATA[travel]]></category>
            <category><![CDATA[trip]]></category>
            <category><![CDATA[meghalaya]]></category>
            <category><![CDATA[prayag]]></category>
            <category><![CDATA[camping]]></category>
            <dc:creator><![CDATA[Prayag Mohanty]]></dc:creator>
            <pubDate>Fri, 30 Oct 2026 20:02:38 GMT</pubDate>
            <atom:updated>2026-10-30T20:02:38.609Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Writing this before I forget any tidbits of the trip forever</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*63nNamrVkXgPF_Jxpa-jYQ.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/960/1*M5PK8xYI3xEDbW1ymzZJFA.jpeg" /></figure><h4>Day 3</h4><p>It was 6 am. I woke up. Quite surprisingly early. Ansh &amp; Rudra were still asleep. This was the perfect scenario for something naughty, I had planned a couple o’days ago. I wanted to scribble on their faces with a marker. I had specifically gotten a marker from the town square last night.</p><p>They were highly restless and light sleepers, which made it difficult for me to draw on their faces. Somehow, I managed to draw a couple of lines on their faces. Ansh thought a fly was on his face and smudged a bit of the line. When he woke up and asked me who had done this to his face, I innocently blamed Sushil. Sushil was flabbergasted and perplexed when Ansh called him out at Eco Park. <br>(In between all this , Ansh had taken a candid pic of me sleeping carelessly)</p><blockquote>The constant banter between Ansh &amp; Sushil is what I live for. It’s just toooo good.</blockquote><p>We grabbed breakfast at the neighboring restaurant, which had aloo paratha &amp; dosas. And their service was so good (pun intended) that tea &amp; coffee came at the end. We went to see the seven sisters waterfalls, but afsoos it was all covered in clouds. Even more afsoos was the fact that Sushil hit his head on the roof of a makeshift stall. Funny af.</p><p>Harshit &amp; Sushil had discovered a fascinating place called Mawmluh caves, which the driver hadn’t heard of. We went there after the letdown that was 7 sisters. The journey was awesome. Windy roads, cool breeze, clear weather, lush green hills like Scotland.</p><p>When we arrived at the entrance to Mawmluh Caves, we couldn’t see the caves. Only an outpost center with a poster of the Mawmluh Caves. Turns out that we needed to pre-book this one, unlike the other walkthrough ones. This was an adventure cave, which had torchlights, ropes and a guide to traverse through. It was closed today anyway.</p><p>So we went to Mawsmai Caves. At the commercial area outside the Caves, there was some banging live music being played. As we entered the caves, there were steps leading right up to the mouth of the cave.</p><p>Nooooo, yesterday’s PTSD kicked right in as we climbed up. Our thighs and calves were cooked. Somehow we reached the mouth of the caves and entered in. The entire experience was very enjoyable. The dimly lit caves and the rocky, wet terrain made me feel like Indiana Jones searching for gold in the caves. We crawled our way through every crevice inside.</p><p>Just near the exit, we took an untrodden path and found a dark, dark cave. There was a hole inside it. As a wise man (me) said, where there’s a hole, there’s a goal. We explored a bit but found nothing. As we exited, it started raining. It was great. A bunch of women had dressed up in the local Khasi attire at the exit.</p><p>I’ve got to say I’ve seen some really beautiful women throughout the trip, but I never really took a chance at connecting with them. I usually do, with the right group. Quite frankly, the boys would get it. Ansh &amp; Sushil were committed already, Rudra was in the talking stage with a few &amp; Harshit was married to his work. So, I had to get it out &amp; see if I still had it in me to charm them. The OG rizzler at work.</p><p>My eye caught a baddie with good fashion sense. She wore a red full-sleeve shirt with jorts &amp; was with her friends. I mustered the courage, went to her &amp; talked to her, complimenting her fashion sense; She and her friends giggled profusely. She told me she was from Manipur. Manipuri baddies do be moving different. Anyway, I left after talking to her, grinning from ear to ear. Feels good to spread happiness. Hahaha. Well, not everything has to be about getting a phone no. or Insta ID (my phone was dead, so it wouldn’t have worked out either). Sometimes, it’s about appreciating others and having a convo.</p><p>Anyway, we headed to the Eco Park in Cherrapunji. I bought a Naga-Khasi pattern shawl for my parents. Ansh and Rudra got ice creams? A brave choice. We whiled away inside the park, which was surrounded by clouds. We played a bit in the children’s playground. There wasn’t much to see, but we had a lot of fun together.</p><p>We then left for the Garden of Caves, which was 3 hours away. It seemed like an eternity, but we finally arrived. The entrance was nondescript. The entrance fee was 100 per person, a steep hike from the 30–50. This place was known to have 11 unique natural spots. Spot no. 9 was the one I was interested in purely for nomenclature reasons coz G is also the 9th letter of the alphabet. I’d have found THE spot. Anyway, the garden had a route with guides planted near every spot, spotted with caves, waterfalls &amp; natural springs throughout. It started raining again in the end, and we took refuge in a stall. We ordered coffee and Maggi. That combo does hit you different in the rainy hill stations of Meghalaya.</p><p>I contacted Lemy, whose camp I had booked at Dawki, to inform him about our arrival and change in contact info. We had lunch at a good dhaba en route. When we left, the outside turned pitch black. The road to Dawki was genuinely scary. No streetlights and low visibility due to fog, kept me awake. Rudra had peacefully slept through this, completely unaware. The pee break in the middle of nowhere in pitch darkness was easily one of the most memorable things outta this trip. There’s a different thrill in being alone in the middle of nowhere in complete silence. A lot of deep talks were had.</p><p>On the bridge to Dawki, we saw the Bangladeshi border and town lights in the distance. After 5 long hours, we finally arrived at Shnongpdeng. Elias, our caretaker, had arranged for dinner at a local kitchen hut. Oh, it was just so homely and tasty!! I gulped a lot.</p><p>I had booked 3 camps, but the camps could only accommodate a few people. The host gave us an extra camp too. It was unusually hot. We settled in and spent the night sharing stories by the river, on the large rocks, which were much cooler. Quite a lovely time. I may or may not have peed directly in the river. I may or may not have troubled Sushil in his sleep.</p><h4>Day 4</h4><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*hTQWmDudHIr9sxwO3wl8KQ.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*0cbcZNATzE4jTMFW5p1L4g.jpeg" /></figure><p>I woke up pretty early. The view of the place was spectacular. The river wasn’t clear but it was amazing. I roamed about the river side, about the rocks, across the river bridge and then refreshed myself. The other woke up slowly, one after the other. Somehow they were sticklers for cleanliness and didn’t prefer to brush/pee/dump at the dirty toilets. I relieved myself for the day completely, I can handle a few minutes of temporary disgust.</p><p>I’ve wanted to have something local for a long time and finally got my hands on jadoh- a local breakfast delight. It was just yummy. Got your belongings together and headed to the Bangladesh border.</p><p>The road was packed with large cargo trucks carrying essentials and goods in and out of the border. Before we realized it, we had reached the border gates. We went ahead and clicked pictures in front of the Bangladesh gate. The Indian gate, called ‘Maitri Dwar ’, was far better equipped than the simple welcome sign of Bangladesh. There was a small obelisk beyond which no one was allowed to cross. We came across a jolly group of Bangladeshis on the opposite end of the border, all wearing their national jerseys. One of them started interacting with us, and I initiated the conversation from my end: “Kemon acho,” in my broken Bengali. We shook hands and right at that moment, a stern warning came from the BSF guard to avoid any physical interactions. Haha it was fun. We bade them goodbye and headed to Dawki river boating.</p><p>When we reached the area, it was rather very uninviting. The river was dried up and the boating looked quite uninteresting, quite frankly. We sat down near a stall and had some snacks. India vs Pak Men’s cricket finals at Asian Games 2026 was happening at the same time &amp; we were watching it. We headed to Mawymslong, which was said to be Asia’s cleanest village. Harshit and Sushil had a heated banter and the bet was on whether India would cross 200 runs or not. Thanks to Tilak Varma, India did &amp; Sushil just lost a thousand bucks to Harshit.</p><p>It was a long road, but it was a nice place when we reached. Every village we went to had a checkpoint with an entry fee. But Mawsynlog was like a flagship village, with a wide variety of flowers planted by the roadside. It was like a well-kept park with some people living inside it. There was a bamboo skywalk that felt like it could topple at any second. We reached the end of the village as we kept walking. We took an Alto taxi on the way back, which was fun. Had a different kind of momo at one of the stalls there &amp; headed straight to Guwahati.</p><p>We stopped on the way for lunch at a nice restaurant, where we gobbled wontons, mutton kheema &amp; whatnot, all very sumptuous. It was a long, long way to Guwahati. We got some assorted Cherrapunji gin bottles from open shops by the road. Ansh was looking for a white variety, the cherry blossom flavor. After stopping at 4 shops, we finally found it at the 5th one!</p><p>Ansh then narrated a harrowing tale of his experience in office politics, and it was wilddd!!</p><p>We finally reached our Guwahati Airbnb, which was located in a residential society. It was a sweet, cozy place. We settled in. Sushil left for Pune, and we ordered dinner. I had ordered a set of pins to remove my card slot and the delivery folks came up to our homes to deliver. Dinner was chichken momos, thukpa and a roll. Damnn good.</p><p>Since all of us were tired and had an early flight the next day, everyone except me dozed off. I spent the entire night editing photos for my Insta stories, sorting pics, and arranging them. And soon it was 4 am.</p><h4>Day 5</h4><p>I slept for 2 hours and packed my suitcase when I woke up. We soon left for the airport, which was hardly 10 minutes away. Oddly enough, the airport charged 200 rupees as a cab entry fee for the departure terminal. We quickly boarded, and at the security check-in, the guard asked me where I was from &amp; I said, IIT Bombay. We then had a 5-minute conversation about his son’s future and career. Though my lads were from BITS (which is equally prestigious imo), me being at IIT always gets a bias in Indian society. Read as tagda dahej (jk!:)</p><p>Surprisingly, as soon as I entered the boarding terminal, the Mumbai flight announced its final boarding call. So I had to bid adieu to the rest &amp; rushed to the gates.</p><p>Overall, it was a great experience. I had a blast reliving the old days with my boys while collecting some new adventures along the way. I’m unsure if I’ll be in India or able to take this trip next year, which makes this experience even more special to me.</p><blockquote>This is a memory, I’ll cherish. Always.</blockquote><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=a012107960b9" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Why does travel cost the way it does?]]></title>
            <link>https://medium.com/@topper1728/the-7-line-d5073df68ed3?source=rss-78ce477380be------2</link>
            <guid isPermaLink="false">https://medium.com/p/d5073df68ed3</guid>
            <category><![CDATA[travel]]></category>
            <category><![CDATA[economics]]></category>
            <category><![CDATA[policy]]></category>
            <category><![CDATA[fuel]]></category>
            <dc:creator><![CDATA[Prayag Mohanty]]></dc:creator>
            <pubDate>Tue, 27 Oct 2026 12:17:47 GMT</pubDate>
            <atom:updated>2026-10-27T13:45:23.537Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Petrol costs roughly the same across urban India, but getting across town does not. This is my analysis of travel cost variance across Indian metros.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*IbdNhZY4DDu0AY1v.png" /><figcaption>I love travelling on the Aqua line metro</figcaption></figure><p>Take a five-kilometer trip in Mumbai. <br>On the local train it costs about ₹5. <br>A BEST bus charges ₹15–20. <br>Drive yourself and the petrol alone costs about ₹7.40/km. <br>Flag down an auto-rickshaw and the meter runs at ₹65.<br>A taxi charges ₹100. <br>That is a 20-fold spread for the same distance in the same city.</p><p>When I recently visited Bhubaneswar, the auto fare seemed a bit too high. This got me wondering about the different fares across India.</p><p>I collected pump prices for 32 states and union territories, CNG prices for 23 cities, regulated auto and taxi fares for ten cities and bus and metro fare tables for the six biggest metros. The common assumption is that fares follow fuel: when petrol goes up, autos and cabs follow.</p><blockquote>Fuel explains almost none of the variation. Who sets the fare, and when they last did, explains most of it.</blockquote><h3>Same crude, different pumps</h3><p>Every pump in India sells fuel refined from crude bought on the same world market, and the central government’s excise duty is the same in every state.</p><blockquote>The differences come from what states add on top: value-added tax, which most states charge as a percentage and so compounds on the base price, plus local cesses and the cost of trucking fuel from the depot.</blockquote><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Z1GWp4QDhzkCzC2a_4Lylg.png" /></figure><p>The result is a clear geography.</p><p><strong>The southern and central states charge the most:</strong> Guntur in Andhra Pradesh sells petrol at ₹117.68 a litre, Hyderabad at ₹116.15, Bhopal at ₹114.54.</p><p><strong>The cheapest pumps are in small union territories and hill and frontier states, where taxes are lighter: </strong>Daman at ₹99.44, Dehradun at ₹99.54, Itanagar at ₹99.64.</p><p>Delhi, at ₹102.12, is the cheapest of the big metros.</p><p>Distance matters within states too. In Andhra Pradesh, Kakinada on the coast pays ₹110.49 while Anantapur in the interior pays ₹118.10, a gap that is likely to reflect freight from coastal depots. Even so, the overall national spread is modest: the costliest state-reference city charges just 1.18 times the price ofthe cheapest.</p><h3>The ₹7 line</h3><p>Convert the pump price into a cost per kilometre for an ordinary car doing 15km a litre and something tidy happens. In all six metros, the fuel cost of driving falls between ₹6.81 and ₹7.74 a kilometre. Draw a line at ₹7 and it splits urban transport almost perfectly in two.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*stxOP7yyBhkX_SAlRHaveg.png" /></figure><p>Everything to the left of the line is public: buses at ₹1.60–3 a kilometre, metros at ₹1–6.40 and Mumbai’s locals at a few paise. Everything to the right is private: autos at ₹11–18 and cabs at ₹12–24. Only two dots sit near the boundary. Delhi’s petrol, at ₹6.81, is the one private cost that falls on the public side, because Delhi taxes petrol more lightly than the other metros. Bengaluru’s proposed metro fare of ₹6.40 for a short trip comes within about a rupee of the cost of driving there.</p><p>The car dot is a floor, not a full cost. Insurance, parking, servicing and depreciation would push it well to the right. A metro that costs nearly as much per kilometre as the petrol for a car is competing on the most flattering comparison the car can offer, and still only just winning.</p><h3>Fuel does not set the fare</h3><p>If fares tracked fuel, cities with dear petrol would have dear autos. They do not. Plot petrol price against the regulated fare and the points scatter with no pattern. The correlation is 0.02 for autos and −0.17 for cabs. Hyderabad has some of the costliest petrol in the country and a middling auto fare. Ahmedabad has some of the cheapest and the highest.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*1e1FK9q9QDBqV6aXrqUimw.png" /></figure><p>The reason is arithmetic. An auto doing 25km a litre burns about ₹4 of petrol a kilometre, which is only 20–38% of the fare. A cab doing 15km a litre burns about ₹7, or 27–60% of its fare. The rest pays the driver, the vehicle loan, the permit and the upkeep. A 10% rise in petrol, passed on in full, would move an auto fare by just 2–4%. Many autos do not run on petrol anyway: Delhi and Mumbai autos run on CNG, which costs ₹86.98 a kilo in Delhi, ₹89 in Mumbai and ₹112 in Hyderabad.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*3EYXJsUPgEWij-bq4tymgg.png" /></figure><p>Chennai’s cab share stands out at 60%. That is not because Chennai’s fuel is expensive. It is because the city’s regulated fare has hardly moved in years, so fuel has grown into a larger slice of a fare that stood still.</p><h3>Markets converge, regulators diverge</h3><p>Line up the spread between the priciest and cheapest city for each item and a pattern emerges. Things priced by oil companies and tax formulas cluster tightly: petrol varies by 1.14 times across the metros and 1.36 times for CNG. Things priced by state transport authorities and metro boards spread much wider: auto fares by 1.82 times, cab fares by 2.08 times and metro fares by 6.4 times.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*SKwSeMXk8nnTPN6UFrhS6w.png" /></figure><p>Fuel prices are set nationally by a handful of companies using a common formula. Fares are set by dozens of state and city authorities, each with its own committee, its own politics and, crucially, its own timetable.</p><h3>Frozen fares fall behind</h3><p>The timetable turns out to be the strongest explanation in the data. Plot each city’s auto fare against the years since its regulator last revised it and the line slopes down. Across ten cities, each year a fare stays frozen is worth about 50 paise a kilometre less. The correlation, −0.61, is far stronger than anything fuel shows, though ten cities is a small sample.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*EAszf-bgNNOSW3qvTF2yBA.png" /></figure><p>Chennai is the extreme case. Its auto fare of ₹25 for the first 1.8km and ₹12 a kilometre after that was set in 2013. Unions want ₹20 a kilometre, and the state held public consultations in September 2026. Hyderabad revised its fares this month for the first time in 12 years, to ₹14 a kilometre. Mumbai sits at the other end because it uses a formula from the Khatua committee that links fares to a cost index, so increases come small and often: the most recent, in September 2026, added about ₹1 a kilometre for autos.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*c9rqwCgYuvrHqPY6V5gTyg.png" /></figure><p>When the official fare falls too far behind, the market finds a way around it. In May 2026, auto drivers’ associations in Ahmedabad raised fares themselves to ₹30 minimum and ₹20 a kilometre, without state approval; the state’s last notified rate, from 2022, was ₹15. In Bengaluru, after Karnataka banned bike taxis, app-booked autos ran 20–30% above the meter fare, according to Deccan Herald. A frozen meter does not keep rides cheap. It moves the real price off the meter and onto apps and haggling.</p><h3>Who pays for the ride</h3><p>Public transport fares follow a different logic: they reflect who carries the cost of the system.</p><p>Mumbai’s suburban railway is the clearest example. <br>Its trains carry nearly 8 million people a day across the metropolitan region, according to the Observer Research Foundation. Western Railway alone logged 25.6 crore journeys between April and June 2026, roughly 28 lakh a day. Packing that many people into each train spreads the cost of running it very thin. But density is only part of the story. Western and Central Railway recover only about 37% of their suburban costs from fares, and Indian Railways covers the gap from its freight profits. Fares are also politically hard to raise: when the railways raised fares in July 2025, suburban tickets and season passes were left untouched.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8GeMkFnxefHbPOycsgZU0g.png" /></figure><p>BEST, Mumbai’s bus operator, recovered about 64% of its costs before May 2025, when it doubled its minimum fare to ₹10. Its passenger revenue had fallen below ₹700 crore a year, and the revision was expected to raise about ₹590 crore more. Bengaluru’s Namma Metro sits at the other extreme. Its projected fare revenue for 2025–26, about ₹1,049 crore, slightly exceeds its projected costs of about ₹1,000 crore.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*W3Vof9VC4ydj1gZDtoDDDw.png" /></figure><p>The cheapest rail fares belong to systems whose debt sits on a national balance sheet. Mumbai’s locals and Kolkata’s metro are both run by Indian Railways and priced like railways: Kolkata’s Line 1 charges ₹5–25 for a whole journey. Delhi and Chennai run as joint companies of the central and state governments and charge in the middle. So does Mumbai’s new underground Aqua Line, which charges ₹10 for up to 3km and ₹20 for 3–12km on the same slab as the city’s other new metro lines: about ₹4 a kilometre for a 5km trip and ₹2 for longer ones. That is cheap for a metro but some 35 times the local train per kilometre on a short trip, and the Aqua Line carried about 1.8 lakh riders a day in April–June 2026 against Western Railway’s 28 lakh. Delhi raised fares in August 2025 for the first time in eight years, to a minimum of ₹11 and a maximum of ₹64.</p><p>Bengaluru uses the same joint-company structure but sizes its fares to service its loans. Its loan repayments are projected to rise from ₹775 crore in 2025–26 to ₹2,348 crore by 2028–29, and the operator works on a rule that metro fares should be about 1.5 times AC bus fares. The 2025 revision raised fares by 51.5% on average but by close to 70% on the 8–15km trips that most commuters make. A further increase planned for February 2026 was put on hold.</p><p>Hyderabad’s metro was built differently. It is a public-private partnership in which L&amp;T holds 90% and the Telangana government 10%, so the fare box has to repay private capital. L&amp;T says the state’s free bus rides for women, the Mahalakshmi scheme, pulled riders off the metro, though there is no authoritative data to check that claim. The state has given L&amp;T a ₹3,000 crore soft loan, and L&amp;T has said it plans to exit after 2026. A fare rise in May 2025 was followed within days by a 10% discount after commuter complaints.</p><h3>Long trips are the bargain</h3><p>Metro fares are charged in distance slabs that rise steeply at first and then flatten, with a cap at the top. So the cost per kilometre falls sharply with trip length. Delhi’s metro costs ₹4.20 a kilometre for a 5km hop but about ₹2 for a 20km commute, and less still beyond 32km, where the fare stops at ₹64.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*6UXsjBn4bL-FzkVdNgcmuA.png" /></figure><p>This shapes who uses what. Metros are priced for long commutes, while short hops are where autos compete. Bengaluru is the exception that proves the rule: even a 20km trip there costs ₹3.70 a kilometre, close to what Delhi charges for 5km.</p><h3>The household arithmetic</h3><p>Turn the per-kilometre numbers into a monthly bill and the stakes become concrete. For a 10km commute each way over 22 working days, the bus costs ₹440–1,320 a month and the metro ₹440–2,332. Petrol alone for a car costs about ₹3,000–3,400. Autos run ₹4,840–8,017 and cabs ₹5,280–10,560.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*oWNUfyR6VoaObYlWRSx69Q.png" /></figure><p>Sharing changes the sums. Divide an auto fare among three riders and in Delhi, Bengaluru, Chennai and Hyderabad each pays about the same as the metro. In Mumbai a full auto would need about five riders to match the metro, and roughly 165 to match the local train. That helps explain why shared autos thrive in so many Indian cities. In Mumbai and Kolkata, where rail is so heavily subsidized, no amount of sharing gets close.</p><h3>What the app charges</h3><p>The regulated fare is only the starting point for most rides now. Under the central government’s 2025 guidelines for ride-hailing apps, the state-notified fare is the base; apps may charge up to twice that at peak times and discount down to half. To see what that means in practice, I checked Uber and Rapido for the same 5km trip in Mumbai, from Aarey on the JVLR towards IIT Bombay, four minutes apart at about 4:50pm on a Tuesday.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*WNhkg2P4nVkl3JmdyiznJQ.png" /></figure><p>Uber’s cabs were priced at almost exactly twice the black-and-yellow meter, and its auto at about 1.5 times. Rapido, which built its business on a flat driver subscription rather than commissions and was described in a 2024 industry analysis as 10–15% cheaper than its rivals, was dearer on nearly everything. Its auto cost ₹198, 48% more than Uber’s ₹134 and 2.2 times the meter. Only its economy cab, at ₹200, undercut Uber, by 5%. Its “bargain” button offered ₹196.</p><p>Several Rapido quotes sit beyond the 2× line. That may reflect taxes, platform fees or time charges in slow evening traffic, and the meter would add waiting charges too. The apps do not break their quotes down, so one screenshot cannot show whether a cap was breached. The “priority” options are clearer: they add 25–40% for a pickup a minute or two sooner, which is surge pricing sold as a feature.</p><p>App rides are also sold by the vehicle, not the seat. Divide each fare across a full vehicle and the ranking flips. A bike carries one person, so it is the dearest app option per head at about ₹23 a kilometre. A full auto, economy cab or XL comes to ₹10–12 per person per kilometre.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*bvTzahbumRYSxQUZo5Ld0A.png" /></figure><p>Even fully shared, no app option crosses back to the public side of the ₹7 line. Mumbai’s metro costs ₹4 a kilometre for a 5km trip and a BEST bus ₹2, the bus five times cheaper than the best shared app ride, and the petrol for a car carrying four works out at under ₹2 a head. This is one trip at one moment, so treat it as an illustration rather than a survey.</p><h3>What would make prices make sense</h3><p>Three changes follow from the data. First, fares should move little and often against a published cost index, as Mumbai does, rather than freezing for a decade and then jumping, which is what happened in Chennai and Hyderabad. Second, if cities want metros to compete with cars, the capital cost cannot sit entirely on the fare box; the systems with the cheapest fares are the ones whose construction debt was carried by the state. Third, the gap between public and private transport is so wide that integrated ticketing and fare caps across bus, metro and rail would matter more to most commuters than any change in the price of petrol.</p><p>The price at the pump is the number Indians watch most closely. For getting around a city, it is one of the least important.</p><h3>How the numbers were built</h3><ul><li>Petrol prices are for 26 October 2026, one reference city per state (Aaj Tak). CNG prices are for October 2026.</li><li>Per-kilometre fuel cost assumes 15km a litre for a car and 25km a litre for an auto.</li><li>Auto and taxi fares are the regulated per-kilometre rate after the minimum fare. App fares and surge pricing are usually higher.</li><li>Bus and metro costs per kilometre are the slab fare divided by distance for the stated trip length.</li><li>Several rates are old or disputed. Bus fares for Bengaluru, Chennai and Kolkata come from 2019 comparisons. Chennai Metro slabs are from 2021, Kolkata Line 1 from 2019 and the Mumbai suburban rate from 2014. Delhi, Kolkata, Chennai, Kochi and Indore taxi or auto fares were last revised in 2023 or earlier. Bengaluru’s metro fare is the proposed February 2026 schedule, which was on hold at last check.</li><li>App quotes are single screenshots taken minutes apart on one trip; per-person figures assume every seat is filled.</li><li>Mumbai Metro uses the shared MMRDA/MMRCL slab (₹10 up to 3km, ₹20 for 3–12km, ₹40 for 18–24km). Line 1 (Versova–Ghatkopar) has its own fare table.</li><li>No figures were found for Hyderabad city buses, Hyderabad cabs or Kolkata autos.</li><li>Correlations use small samples (7–10 cities) and are suggestive only.</li></ul><h3>Sources</h3><ul><li><a href="https://www.aajtak.in/fuel-price/petrol-price-in-new-delhi-today">Aaj Tak, city petrol prices</a> · <a href="https://www.aajtak.in/fuel-price/cng-price-in-ahmedabad-today">Aaj Tak, CNG prices</a> · <a href="https://www.goodreturns.in/news/cng-png-prices-in-india-for-october-2026-check-rates-in-delhi-mumbai-chennai-kolkata-more-cities-1538501.html">Goodreturns, CNG October 2026</a></li><li><a href="https://www.newsbytesapp.com/news/india/auto-and-taxi-fares-hiked-in-mumbai/story">NewsBytes, Mumbai fares</a> · <a href="https://www.uniindia.com/south-india/transport-telangana-auto-fares/200699">UNI, Telangana auto fares</a> · <a href="https://www.deccanherald.com/amp/story/india%2Fkarnataka%2Fbengaluru%2Fbengaluru-auto-drivers-must-recalibrate-fare-meters-by-october-31-3660013">Deccan Herald, Bengaluru autos</a> · <a href="https://www.deccanherald.com/india/karnataka/bengaluru/act-against-overcharging-autorickshaw-drivers-minister-directs-transport-commissioner-3607380">Deccan Herald, app autos after bike-taxi ban</a></li><li><a href="https://www.businesstoday.in/latest/story/auto-and-taxi-fares-hiked-in-delhi-check-new-rates-359828-2023-01-11">Business Today, Delhi fares</a> · <a href="https://www.dtnext.in/news/tamilnadu/tamil-nadu-auto-fare-hike-public-consultations-across-districts-on-sept-23">DT Next, Tamil Nadu autos</a> · <a href="https://www.patrika.com/ahmedabad-news/ahmedabad-auto-rickshaw-fares-hiked-20623243">Patrika, Ahmedabad autos</a> · <a href="https://www.tribuneindia.com/news/chandigarh/shell-out-more-for-hiring-taxi-auto-ut-revises-rates/amp">Tribune, Chandigarh fares</a> · <a href="https://www.freepressjournal.in/indore/taxi-auto-fares-hiked-in-the-city">Free Press Journal, Indore fares</a> · <a href="https://onmanorama.com/news/kerala/2022/05/01/hike-in-bus-auto-taxi-fares-kerala.html">Onmanorama, Kerala fares</a> · <a href="https://wbxpress.com/taxi-fares-2018/">WBXpress, Kolkata taxis</a> · <a href="https://stgcms.storyboard18.com/?p=21857">Storyboard18, Karnataka cab fares</a></li><li><a href="https://studycafe.in/dmrc-announces-minimal-hike-in-delhi-metro-fare-from-august-25-check-new-revised-rates-391356.html">StudyCafe, Delhi Metro fares</a> · <a href="https://www.siasat.com/hyderabad-metro-fares-revised-new-rates-effective-from-may-17-3221038/">Siasat, Hyderabad Metro fares</a> · <a href="https://www.trayaan.com/2026/02/bengaluru-metro-train-fares-to-be-hiked-from-9-february-2026-details.html">Trayaan, Bengaluru Metro fares</a> · <a href="https://www.thenewsminute.com/article/chennai-metro-fares-slashed-february-22-full-details-143818">The News Minute, Chennai Metro fares</a> · <a href="https://themetrorailguy.com/kolkata-metro-information-map-updates/">The Metro Rail Guy, Kolkata Metro</a></li><li><a href="https://www.theweek.in/wire-updates/national/2025/05/08/bes25-mh-ld-best-fare.html">The Week, BEST fares</a> · <a href="https://www.freepressjournal.in/amp/mumbai/mumbai-news-best-doubles-minimum-bus-fares-from-may-9-non-ac-starts-at-10-ac-at-12-toll-and-pass-revisions-announced">Free Press Journal, BEST revision</a> · <a href="https://www.delhitourism.gov.in/transport/city_bus.html">Delhi Tourism, DTC fares</a> · <a href="https://citizenmatters.in/bengaluru-bmtc-bus-fares-high-indian-cities-ridership-public-transport-opinion/">Citizen Matters, bus fare comparison</a></li><li><a href="https://www.orfonline.org/public/uploads/posts/pdf/20230725135913.pdf">ORF, Mumbai suburban rail and BEST finances</a> · <a href="https://www.freepressjournal.in/mumbai/mumbai-suburban-railway-loses-4465-lakh-riders-in-q1-as-metro-gains-on-parallel-routes">Free Press Journal, Mumbai ridership</a> · <a href="https://www.latestly.com/india/information/indian-railways-fare-hike-will-mumbai-local-train-ticket-prices-increase-from-july-01-check-new-ticket-slabs-tatkal-reservation-changes-and-other-key-details-6963709.html/amp">LatestLY, suburban fares spared</a> · <a href="https://techenclave.com/threads/mumbai-has-the-cheapest-railway-fares-in-the-world.168230">TechEnclave, Mumbai local per-km rate</a></li><li><a href="https://theprint.in/india/bengaluru-metro-unlikely-to-revise-high-fares-but-why-is-namma-metro-so-expensive/2779070/">ThePrint, why Namma Metro is expensive</a> · <a href="https://swarajyamag.com/amp/story/infrastructure%2Fexplained-why-lt-wants-out-of-hyderabad-metro-and-the-road-ahead">Swarajya, L&amp;T and Hyderabad Metro</a> · <a href="https://www.businesstoday.in/amp/industry/infra/story/lt-plans-to-exit-hyderabad-metro-post-2026-r-shankar-raman-says-free-bus-rides-have-made-ridership-less-interesting-429059-2024-05-10">Business Today, L&amp;T exit plan</a></li><li><a href="https://www.theweek.in/news/biz-tech/2025/07/02/ride-hailing-goodbye-to-3x-4x-surge-pricing-and-ride-cancellations-as-morth-announced-new-guidelines-for-uber-ola-rapido-cab-aggregators.amp.html">The Week, aggregator guidelines 2025</a> · <a href="https://navbharatlive.com/maharashtra/mumbai/bike-taxi-will-start-in-mumbai-ola-uber-rapido-got-green-signal-fare-for-1-5-km-is-15-rupees-1356661.html">Navbharat Live, Maharashtra bike-taxi fares</a> · <a href="https://www.equentis.com/blog/is-rapido-set-to-outpace-ola-uber/">Equentis, Rapido market share and pricing</a> · Uber and Rapido quotes: author’s screenshots, 27 October 2026.</li><li><a href="https://www.freepressjournal.in/amp/mumbai/mumbai-metro-3-final-stretch-between-worli-to-cuffe-parade-set-for-inauguration-on-oct-8-all-about-citys-1st-underground-line-stations-fare-more">Free Press Journal, Aqua Line fares</a> · <a href="https://www.freepressjournal.in/topnews/mumbai-pocket-friendly-metro-fares">Free Press Journal, MMRDA metro fare slabs</a></li><li>Map shapes: <a href="https://github.com/VictorCazanave/svg-maps">svg-maps/india</a> by Victor Cazanave, CC BY 4.0.</li></ul><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=d5073df68ed3" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Inaugural IIT Bombay Quiz League]]></title>
            <link>https://medium.com/@topper1728/3-75-points-iit-bombay-quiz-league-8d0404851a82?source=rss-78ce477380be------2</link>
            <guid isPermaLink="false">https://medium.com/p/8d0404851a82</guid>
            <category><![CDATA[trivia]]></category>
            <category><![CDATA[iit]]></category>
            <category><![CDATA[college]]></category>
            <category><![CDATA[league]]></category>
            <category><![CDATA[quizzing]]></category>
            <dc:creator><![CDATA[Prayag Mohanty]]></dc:creator>
            <pubDate>Mon, 26 Oct 2026 23:11:26 GMT</pubDate>
            <atom:updated>2026-11-02T13:37:15.463Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Four factions, four quizzes, one SBT win and a league lost so, so narrowly</em></p><p>October at IIT Bombay came with a new kind of quiz league. Instead of teams just turning up and playing, every quizzer belonged to a <strong>faction</strong>, and every quiz fed into a faction leaderboard. Four factions, each led by a pair of captains:</p><ul><li><strong>InstiGators (Insti)</strong>, led by Gayatri and Shriram</li><li><strong>It’s Always Sunny in Powai (Sunny)</strong>, led by Madhav and Basu</li><li><strong>All Seeing Eyes (ASE)</strong>, led by Soham and Johan</li><li><strong>Gunners of Yore (GoY)</strong>, led by Chinmay and Mrunal</li></ul><p>I was a Gunner. So were Devraj, Shagun, Shreyas, and Avaneesh. Banger team.</p><p>The format was simple: each quiz was hosted by one faction’s captains, the top ten teams in the finals earned points for their faction (25 for first, then 18, 15, 12, 10 and so on), with participation points on top. Four quizzes, one table at the end.</p><h3>Quiz 1: India Quiz (6 October)</h3><p><em>Hosted by Madhav &amp; Basu</em></p><p>The league opened with an India Quiz, and I missed it entirely. My weekly presentation was at the same time, and the presentation won.</p><p><strong><em>ASE</em></strong> made the strongest start: Abhineet, Aparna and Nathan took first, and Uttirno, Soham and Arnav came third. <strong><em>InstiGators</em></strong>’ Shriram, Dhruvi and Sunaina were second. For the <strong><em>GoY</em></strong> , Devraj, Gaurav and Sabarish finished fifth, a quiet but useful start.</p><h3>Quiz 2: METH (14 October)</h3><p><em>Hosted by Chinmay &amp; Mrunal</em></p><p>Our own captains hosted the second quiz. I didn’t make the finals this time, but the faction did not need me: <strong>Shreyas, Ching and Arjun won METH for <em>GoY</em></strong>, ahead of two It’s <strong><em>Sunny </em></strong>teams (Zorawar, Kartik and Mihir second; Akif, Aaroh and Aritro third).</p><p>It was also the first sign of who our real rivals would be. Sunny in Powai put three teams into the top five.</p><h3>Quiz 3: SpEnt (16 October)</h3><p><em>Hosted by Gayatri &amp; Shriram</em></p><p>Sports and Entertainment. Shagun and I squeezed into the finals as the 11th team, and finished exactly there: 11th, one place outside the points.</p><p>I’ll own this one. An urgent BBQL quiz landed in between, and I could have done better without it. Meanwhile, our captains Mrunal and Chinmay came third, and Devraj added a fifth place. Up top, <strong><em>Sunny </em></strong>won again, this time through Basu.</p><h3>Quiz 4: SBT (21 October)</h3><p><em>Hosted by Johan &amp; Soham</em></p><p>Then came SBT, my favorite of the lot. Johan blabbered earlier about tailoring the SBT set such that I would never win. Bunch of fluff. Chinmay, Pranav and I <strong>won it by a mile</strong>, finishing ahead of Akif and Kartik Pathak in second.</p><p>On any other day, this would’ve been a great result for the whole faction: Mrunal, Bimal and Arjun came fourth, and Aarnav, Avaneesh and Shagun fifth. Three Gunners teams in the top five, 47 points from the finals alone.</p><p>But alas, it wasn’t meant to be. We won the battle but lost the war.</p><h3>The final table</h3><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*Pj069xOth6nxlqCZ40BgeA.jpeg" /><figcaption>Final faction standings after four quizzes</figcaption></figure><p><strong><em>GoY </em></strong>lost the league by <strong>3.75 points</strong>. <br>Ahhhh</p><p>That is less than a single fifth-place finish is worth (10 points).<br>Less than participation points (5 points)<br>Maybe we could’ve improved our positions instead of takingthose negs. <br>Maybe if Shreyas or Devraj could’ve teamed up. <br>Lots of ifs and buts. <br>It was by the finest of margins that we lost.</p><blockquote><em>One more Gunners team in the top five at any quiz, the final result would’ve looked completely different.</em></blockquote><h3>By the numbers</h3><p>Once the league was over I put the results sheet through a few charts. Some of it confirmed what it felt like from the inside. Some of it didn’t.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*yNWZg3CL0jaDV65025yFbw.jpeg" /><figcaption>Cumulative faction points after each quiz.</figcaption></figure><p>The lead changed hands only once.</p><p><strong><em>ASE </em></strong>opened with 49 points from the India Quiz, almost all of it from two podium teams, and then took 29 points across the next three quizzes combined. <strong><em>Sunny </em></strong>started that first night in last place, went top after METH, and never came down.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*jZwKtzt6L00qk3upEgj-6Q.jpeg" /><figcaption>Gunners of Yore minus It’s Always Sunny in Powai, after each quiz.</figcaption></figure><p>We were actually ahead of <strong><em>Sunny </em></strong>after the India Quiz, by 2.25. METH swung the gap 12 points the other way, SpEnt pushed it out to 17.75, and SBT, where <strong><em>GoY </em></strong>took 47 points to <strong><em>Sunny’s </em></strong>33, pulled it back to 3.75. One quiz short.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*daibqJTzRulVyLId7m-Ryg.jpeg" /><figcaption>Finals points (solid) and participation points (hatched) for each faction.</figcaption></figure><p>This is the one I didn’t expect.</p><blockquote><strong>Gunners </strong>scored more finals points than <strong>Sunny</strong>: 127 to 125.75.</blockquote><p>The league was decided by participation points, where Sunny collected 15 to our 10. We picked up none at SpEnt or SBT, and that 5-point difference is bigger than the final margin.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*5x4dy2m1GXcj5tyqiZ1m1A.jpeg" /><figcaption>Points each faction took from each quiz</figcaption></figure><p>Every faction had one big night:</p><p><strong><em>All Seeing Eyes</em></strong>’ India Quiz (49)<br><strong><em>Sunny’s</em></strong> METH (48), <br><strong><em>GoY’s </em></strong>SBT (47) <br><strong><em>InstiGators’ </em></strong>METH (31).</p><p>What separated the top two from the rest was the floor. After India, Sunny never took fewer than 33 points from a quiz, and our lowest after India was 31, at SpEnt. All Seeing Eyes took a single point at METH.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*92W6rnnxn8zBT4qRlCQqtQ.jpeg" /><figcaption>The faction of every team in every finals slot.</figcaption></figure><p><strong><em>Gunners </em></strong>won two of the four quizzes, more than any other faction, and our finals teams had the <strong>best average finish (4.9th) and the most points per finals appearance (11.5)</strong>.</p><p><strong><em>Sunny </em></strong>won through depth instead: 12 finals slots, the most of anyone, and four podiums. Both factions finished with seven top-five places.</p><p>InstiGators reached the podium three times without winning a quiz.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*nYdlGkF0nEbh8AOFrVHmIg.jpeg" /><figcaption>Finals points won by each quizzer’s teams, with full team points credited to every member</figcaption></figure><p>Credit every member with their team’s points and <strong><em>Sunny’s </em></strong>depth shows up again: Kartik (50), Akif (47.5) and Debayan (47) top the list, each reaching three or four finals. Soumya (41) carried <strong><em>InstiGators</em></strong>. For the Gunners, Chinmay (40) leads with just two finals: the SpEnt podium and the SBT win.</p><p>That is what made the league fun. Every finals slot counted, the faction rivalry was real, and nobody, including me, could relax at any quiz.</p><p>Sucks to lose the overall league even after topping SBT. But ig that’slife. Congratulations to Madhav, Basu &amp; KP. Thanks to all the captains &amp; quizzers for running it.</p><p><em>More quiz sets and updates: </em><a href="https://prayag-mohanty.github.io/QM/askaban/"><em>QM Prayag</em></a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=8d0404851a82" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Da Double Decker Dilemma]]></title>
            <link>https://medium.com/@topper1728/the-double-decker-dilemma-cec304265457?source=rss-78ce477380be------2</link>
            <guid isPermaLink="false">https://medium.com/p/cec304265457</guid>
            <category><![CDATA[prayag]]></category>
            <category><![CDATA[meghalaya]]></category>
            <category><![CDATA[root-bridge]]></category>
            <category><![CDATA[cherrapunji]]></category>
            <category><![CDATA[travel]]></category>
            <dc:creator><![CDATA[Prayag Mohanty]]></dc:creator>
            <pubDate>Sun, 11 Oct 2026 12:34:18 GMT</pubDate>
            <atom:updated>2026-10-26T20:12:31.715Z</atom:updated>
            <content:encoded><![CDATA[<p><em>I’m not sure how we survived this is PTSD-inducing trek. Definitely not amateurish by any metric.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1000/1*mgFhdGJqGGuCk51sURs3sQ.gif" /></figure><h3>Day 2 (1 Oct): The trek</h3><p>I woke up at 6 (slept at 2 am) to the beautiful view of Tyrna, with an azure-blue sky &amp; lush green hilltops wrapped in silky, misty, bright white clouds. The temperature was cool, not too cold, not freezing cold either.</p><p>Today’s plan was to hike all the way to the Double Decker Living Root Bridge &amp; then later to another waterfall further on the way. The whole journey was said to take 6–8 hours back &amp; forth.</p><p>I was very, very excited because I had heard &amp; studied a lot about these living root bridges, and also because this was the main motivation for the trip to Meghalaya.</p><p>The selection of Smoky Falls resort was also precisely due to its proximity to the starting point of the Double Decker Living Root Bridge trek.</p><p>We reached the starting point at 7:15 am. Coming into the trip, Harshit was down with a fever. So, he wore a standout red jacket at all times. Probably indicating his type in women as well. Ansh got us a guide for the trek. Sushil &amp; I got ourselves bamboo trekking sticks, primarily because they fit the trekker’s look, but, more importantly, we were sure we would struggle midway &amp; this stick should help us, even if temporarily.</p><h3>Downhill is a scam</h3><p>And so it began.</p><p>Interestingly, the trek starts at the top of a hill as a 3 km downhill stretch, then continues for another 3 km across reinforced bridges to the living root bridges.</p><p>We started off on a good, enthusiastic foot. The guide’s name was Kitbo &amp; he had a doggo friend who, believe it or not, walked alongside us the entire trip, end to end. He wore chappals, compared to the shoes &amp; dri-fit we wore. The way the dog traversed the entire trek reminded me of the Mahabharata story of the dog that traveled to the heavens with the Pandavas (we were 5 as well).</p><p>The downhill trek was essentially a staircase route with village homes on either side. The route was lined with stalls made of corrugated steel popping up every few metres, selling snacks, juice &amp; tea items.</p><p>Contrary to popular belief, a downhill walk is not easy. The incline was too steep. My legs shook &amp; shivered on every step I took.</p><p>ON GOD, we had to take 2 breaks before we reached the valley. Harshit had to take a pee break, so we sat down at a stall nearby and sipped some lime juice. The next half of the trek involved river crossings.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*kHWIXB1uAzSko3giK6F3Jg.jpeg" /></figure><p>We traversed 2 long man-made suspension bridges. Although very shaky, it was fun &amp; beautiful. Especially the scenery around. Within no time, we reached the famed root bridges. It was relatively less crowded, so we were able to take a lot of photos &amp; chill for a bit.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*svv4AEewCZnEQklgB_ABNg.jpeg" /></figure><p>It was exactly how I imagined it to be. I was surprised at how strong the bridge was naturally. There was a variety of tourists that I could spot: couples on their honeymoon, Bengali uncs who’ve somehow got their hands on Insta pro and are vlogging, influencers with a huge-ass setup &amp; camera angles, French tourists, boy bunches like us.</p><p>At this point, we could either go ahead to see &amp; bathe at the falls or head back. We took a poll on who wanted what. Sushil gave up instantly &amp; wanted to head back. No surprises here, Harshit also wanted to go back.</p><h3>Look, don’t touch</h3><p>So, it was me, Ansh &amp; Rudra who decided to go ahead. The next 3 km was a worthwhile “trek” trek, with a mix of uphill &amp; downhill terrain, with sometimes permanent, sometimes kucha rocky paths. It was long &amp; we did take 3 breaks, but it was defo. easier than what was about to come.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*iMfvB0be15t374PfAClWFg.jpeg" /></figure><p>Lo and behold, the Rainbow waterfalls appeared right in front of our eyes. Ansh &amp; I had continued the trek with the expectation of bathing in the cool, cool pool of the waterfalls. But on reaching the spot, we were informed that we could not go closer to the waterfalls &amp; could only see them from afar.</p><p>Aah fuck. This was damn sad. However, Kitbo assured us he’d take us to a secluded pool where we could take a nice dip &amp; relax. But for the moment, we sat down at the viewing point of the Rainbow Falls &amp; rested with the breeze hitting our faces. While we were sitting there, something magical suddenly happened, revealing to us why the waterfall was named that.</p><p>For a moment in between, the sun shone brightly through the mist generated by the waterfall&#39;s impact on the giant boulder beneath it. That created a rainbow outta nowhere &amp; whatta sight it was for sore eyes!</p><p>I met an influencer couple, who caught my eye as they were flying a DJI Mavic to capture the aerial view of the falls. A bunch of shirtless Mallu dudes were looksmaxxing and posemaxxing in front of their girlfriends, with the waterfalls as the backdrop.</p><p>The way back to Nongriat was chill. Sure, we took 1 break.</p><p>During that break, Rudra said I was hot. I thanked him for his compliment. Tell me something I don’t know…</p><p>But he insisted that I was actually emanating smoke. My body was so smoking hot that it was producing fumes. Even Ansh agreed.</p><p>I couldn’t believe it. I removed my t-shirt &amp; started recording myself. Holy smokes, that’s true. Probably due to the lower temperature and low air pressure, my body felt like it was sizzling. Damn!</p><p>I then decided to trek shirtless to cool off &amp; look cool, too. Why should I deny others the pleasure of seeing my hot bod, I reckoned.</p><p>Don’t blame me if someone appreciates it or falls in love with my superficial self. Win-win, I’d say!</p><p>Anyway, after stalling our plans to bathe, we finally arrived at a pool inside a slow-flowing river under the 1st bridge we crossed!</p><p>The stones were hot, but the moment we (me &amp; Ansh) took a dip in the pool, our muscles relaxed immediately. Rudra wouldn’t come in coz he didn’t have spare clothes to change into. Ansh couldn’t swim, so I guided him at times. There were tiny fish that bit off dead skin and gave a tingling feeling. I enjoyed it.</p><p>I had brought a “waterproof phone case” with me in case I wanted to record videos in water. Easily my biggest mistake of the trip. Becoz after I put my phone in it and took it into the pool, the network connectivity, data and camera stopped functioning. All my UPI transactions, WA calls, and texts were cut off immediately. But thankfully, my friends lent me their phones to take pics or make calls if I needed to anytime during the trip. Especially Rudra, who was always patient and helpful.</p><p>While I used the good old towel method to change clothes, Ansh had a hard time finding a secluded spot to change his; thankfully, he found one.</p><p>My mind and body were refreshed. But not for long.</p><h3>3500 steps. My foot.</h3><p>We reached the foot of the mountain. We were not ready for what was about to hit us. The steep stairs went up and up and I could not see their end. I kept on going, yet the steps wouldn’t finish. People sat on the stairs to rest, drank water and tried to bring themselves to climb. We did too.</p><p>3500 stone steps, apparently. My foot.<br>I was out after every 100 steps.</p><p>Mind was willing, but my body couldn’t carry on. My stamina was depleting. My huffs were growing heavier. Stops were becoming frequent. Resources ran scarce, as I was out of water, out of food. No money atp.</p><p>I genuinely felt that we might not make it out today. There were locals with bamboo stretchers at a few points, who would carry you out to the top for ₹8–12k. It was a mental hurdle, and I was pushing it hard.</p><p>We must’ve taken at least 7 breaks on the way back to the top. Kitbo kept encouraging us city dwellers that the end was near, lying through his teeth to keep our spirits high (My end was definitely near) I gave up my stick to Ansh as he needed it more than I did. Rudra &amp; Ansh kept asking for more breaks, but I insisted they move on. I realized the more breaks I took, the more time it took to cover, and the longer the overarching mental sword hung over my head.</p><p>Ab mazaa nahi aa rha tha. It was no longer fun. It was pure torture. For the final stretch, I kept going and going until I saw the turn that led to the entrance of this whole nightmare.</p><p>I shouted and almost cried out of joy. FINALYY!!</p><p>Right by the entrance, a cute lil girl with a lovely voice was selling fruit, juice, and snacks. I bought some mandarins, and I kid you not, to my parched, broken body, they tasted like heaven. It was as if my body regained life after drinking an elixir. It was the best thing ever.</p><h3>Sohra mai Qohra</h3><p>Once we returned to the resort, we feasted. The food was delicious. Same fare as last night, plus chicken fried rice. We quickly went on to our stay at Cherrapunji. On the road, things slowly became chillier and foggier.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*RpqzI2_GG6Potumm.jpeg" /></figure><p>Noble Hotel. It was located right at the Sohra town square. It was also where Sushil &amp; Harshit had already settled &amp; slept after their return from the Living Roots Bridge. We convened at a Khasi shop for tea, which had school-length tables and benches.</p><p>When Harshit asked how the trek was, all three laughed it out to cope and described the horror of the uphill trek. Tbh, we were doing absolutely fine until the uphill climb. Harshit resonated with us and said their legs had completely given up by the end of it. On top of that, Sushil struggled too. He gave his bag and stick to Harshit to carry for him. Harshit was sick already and had almost collapsed.</p><p>Massive respect. So much to stay in Sushil’s good books.</p><p>Luckily, I had brought my laptop, so I informed my parents about the phone. While the rest slept or relaxed, I roamed about the shops of the town square. All the shops had 1 LED bulb, so the lighting was dim. The sun sets early, so they all close early. All establishments were closed by 9 pm. Even our own hotel’s restaurant.</p><h3>A ₹9k dinner and ₹87 in winnings</h3><p>Ansh recommended going to this resort called Polo Orchid for dinner. Heard it was highly rated. Turns out it was a proper luxury resort with a classy restaurant. There was good live music. The look &amp; feel was amazing. The vibes were immaculate.</p><p>We ordered a lot and finished everything. Their service was good. They remade the biryani just coz I found it too bland. Highly recommend. The bill came close to ₹9k. Lol</p><p>Sushil got a call from the hotel asking us to be back by 10:30 pm, as the gates close by then. What a strange town. Also, Rantu Dutta must’ve had a couple of pegs in the parking lot becoz he bumped the back of his car into a garden fence and forgot to pull his handbrake when he went to check it.</p><p>Anyway, we returned to our hotel. I took a bath and then played poker with the gang. ₹150 buy-in for 300 chips. Harshit was on a rampage, guzzling chips from everyone and won around ₹400. As an absolute noob at poker, I somehow managed to hold my own &amp; won ₹87. Better than last time for sure.</p><p>Post poker, we went to our rooms and retired for the night.</p><p><em>Coming up next, Day 3: Caves, flirting around, treacherous roads &amp; camping</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=cec304265457" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[(Dandi) March to Meghalaya]]></title>
            <link>https://medium.com/@topper1728/dandi-march-to-meghalaya-7f3a066198eb?source=rss-78ce477380be------2</link>
            <guid isPermaLink="false">https://medium.com/p/7f3a066198eb</guid>
            <category><![CDATA[holidays]]></category>
            <category><![CDATA[shillong]]></category>
            <category><![CDATA[meghalaya]]></category>
            <category><![CDATA[prayag]]></category>
            <category><![CDATA[travel]]></category>
            <dc:creator><![CDATA[Prayag Mohanty]]></dc:creator>
            <pubDate>Sun, 11 Oct 2026 02:33:39 GMT</pubDate>
            <atom:updated>2026-10-11T05:34:07.616Z</atom:updated>
            <content:encoded><![CDATA[<p><em>5 friends, 4 days, 3 places, 2 rooms, 1 car and an absolutely awesome trip.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*23exCz9ubzFtF3jIw5PT3w.jpeg" /><figcaption>At some point on the second day, I walked out of our room and found myself standing inside a cloud. Not a metaphor. An actual cloud. This is how I got there.</figcaption></figure><p>The North-East has always been an exotic, almost-fantasy place to me. Untouched, rarely covered in mainstream media. Always somewhere I’d get to “someday”. It has eluded me for a very long time.</p><p>So when Ansh asked me whether I’d be interested in a boys’ trip over there, it was hard to say no. Anything for the plot. YOLO.</p><p>Despite my MTP work, PhD applications &amp; the related anxiety, I jumped at the opportunity. The initial plan was to visit Sikkim, but since I had already been there, Ansh agreed to shift gears to Shillong &amp; around.</p><p>I must say that it has become a tradition to travel to a new destination every 2nd October weekend (Ty Baapu!) with ma homies, Ansh, Sushil &amp; Harshit. It was Alleppey last year.</p><p>The WA group was aptly called ‘Dandi March,’ alluding to Gandhi’s iconic walk.</p><p>And it is Meghalaya this year, with Rudra joining in as well. I love Rudra. Such a ball of joy, a smart, awesome junior who also happens to be amazing at badminton. What a guy (kya banda hai!).</p><p>Somehow, none of them are in touch with me through the entire year &amp; suddenly, we catch up &amp; divulge everything in the next 3–4 days.</p><p>All of us finalized Meghalaya in July. We decided on the itinerary &amp; zeroed in on Shillong, Cherrapunji, Dawki &amp; Guwahati.</p><p>But we stalled booking tickets till August and even longer in confirming our accommodation till September. Probably because everyone was busy with their own thing or something else. I’m still not sure why we delayed the bookings.</p><p>The group chat was pretty much dead until T-5, when the boys locked in, excited for the trip ahead.</p><h3>Day 0 (29 Sep): 5 am flight</h3><p>I was overloaded. An MTP presentation I’d finished only the day before, plus a lab for my public policy elective (PS 648, Data Visualization for Public Policy). Clothes washed and ironed, shoes cleaned butnothing packed.</p><p>Sushil was coming from Pune &amp; landed at 5:40 am. Rudra joined me (from Goa) to depart from Mumbai. Harshit &amp; Ansh came from Bengaluru. All 4 of us landed at around 8:00 am. My flight was at 5 am.</p><p>Not to forget, that night the IITB badminton team had a pre-camp dinner. It is an informal tradition to kickstart the pre-camp training for the Inter IIT Championship! From 6–9 pm, I had the public policy lab. 9–12 am, pre-camp dinner.</p><p>Here’s where it gets stupid. I had already started packing &amp; my jeans were tucked away, so I ended up wearing shorts &amp; Crocs to the restaurant. Upon arriving, they denied me entry due to the dress code. Somehow, Sai &amp; I convinced the lady &amp; her manager to let us in (where we exited after 10 min, coz the vibes were lowkey dead &amp; the music was LOUD asf).</p><p>Anyway, I left the dinner party at 12:30 &amp; by the time I started packing, Rudra had already landed at T2. I guess that’s the advantage of living in Powai, especially at night, the airport is 20 minutes away. Thanks to my overthinking, I finally finished packing by 2:30. Left at 2:45 &amp; reached T1 at 3:15.</p><p>To you, it may seem like odd hours; it was another Wednesday for a nocturnal beast like me. Rawr.</p><p>T1 is pretty compact. It hardly takes a few minutes from check-in to the boarding gates. Small, petite. I love it.</p><p>Right after the security check, I found a lost soul wandering around the security area.</p><blockquote>“Arre Rudra!”</blockquote><blockquote>We hugged &amp; exchanged pleasantries. He looked slightly perplexed.</blockquote><blockquote>“Maine shayad meri hostel keys idhar chod di thi, trying to find it.”</blockquote><blockquote>“CISF help desk pe puch, unko pta hoga.”</blockquote><p>Luckily, the helpdesk had his key &amp; handed it over to him.</p><p>We sat for a few minutes before the boarding call. Since we had checked in separately, we were seated separately.</p><p>It was here that I got to know that Rudra had to take a slight detour from the trip, as he had an online interview scheduled the day we landed. His dad had booked a Novotel for him.</p><p>I slept through the entire flight. I do not recollect anything except for the magnificent gradient of the sunrise on the horizon. I’ve never seen any gradient as cool as that one. Fucking beautiful.</p><p>Sushil was already waiting, &amp; Ansh &amp; Harshit joined in shortly. Rudra had left for the hotel by then. We had the entire day to ourselves. Lessgooo!</p><h3>Day 1 (30 Sep): “Who’s S*****?”</h3><p>Ansh had booked a cab driver with an Innova Crysta via his contacts. Very resourceful guy who drinks at every opportunity.</p><p>Anywho, it was a lovely car, &amp; Rantu bhaiya was the man behind the wheel, who called himself Romtu Datta due to his heavy Assamese accent. We had plenty of misunderstandings in our conversations with him during the entire trip. Interestingly, throughout the trip, wherever we stayed or ate, the establishment also provided for the driver’s meals &amp; accommodation. Everywhere in Meghalaya.</p><p>He was from Jorhat &amp; was a busy, busy driver during the on-season. His ‘maalik’ took care of him in the off-season.</p><blockquote>While talking to him, I told “Dowki Nodi” in an accent</blockquote><blockquote>Ansh couldn’t stop laughing when I said this.</blockquote><p>Anyway, back to the trip.</p><p>The reason you travel to new places is to experience new things and eat different food. Right? Yaar, the first place we stopped for breakfast was Sardar Dhaba. Guwahati mai hoke dosa aur aloo paratha pel rahe hai instead of having something local.</p><p>Ansh shared an interesting fact: the IND vs WI 2nd ODI was taking place at the Guwahati stadium that very day. Whatta coincidence!</p><p>But would I want to spend any time indoors, inside a stadium, for a match I wouldn’t have bothered about until it was told to me rn, or would I spend that time frolicking around the open pastures of Shillong?</p><p>It was obvious.</p><p>I’m not as much of a cricket fan as the other 4. More of an F1, badminton, football, and tennis fan. I usually become one during the WC or IPL.</p><p>It was a long route to Shillong, so I started playing some music on the car stereo. Anyway, we started filling each other in on our lives! (It was far, far from what I expected.)</p><p>Sushil kept repeating one name &amp; kept bringing it back again &amp; again in our conversations.</p><p>S***** studied in Guwahati. S***** did this, did that.<br>Something didn’t add up. Who’s this person I wasn&#39;t aware of that these guys know about? Am I missing something?</p><blockquote>“Dawg, who’s this S*****?”</blockquote><blockquote>“Oh! I haven’t told you kya? S***** is my girlfriend.”</blockquote><p>I went into denial. Not because Sushil had a gf (I mean, he is perfect bf material) but because I thought he was the one TRUE love of my life.</p><p>They were 10–11 months strong.</p><p>How could you not inform your best friend about this?</p><p>Perhaps this trip was how he wanted to inform me.</p><p>Nonetheless, I was happy for my homeboy.</p><p>There are wild stories, and then there are Harshit’s stories. The things that happen with Harshit are wilder than your dark fantasy.</p><p>AI can never come up with the plot twists Harshit cooks up. One of the stories he told had to do with his low-hanging fruits and, let’s just say, the story was pretty swell.</p><p>Another story of his involved his flatmate. In this story, everyone was drunk. His flatmate got a girl home &amp; a few hours later, the girl’s husband came to the flat in utter rage.</p><blockquote>“You could either watch it happen or join &amp; be part of it” — Elon Musk</blockquote><p>Quite a scene.</p><h3>A lake, a bad thali and an elephant that isn’t there</h3><p>In the midst of all these intellectual talks, we reached the expansive Umiam Lake. The calmness of the lake, the green hills surrounding it, the forests around: it all looked very picturesque….but boy, the heat was killing us.</p><p>We quickly got on a boat ride on the lake. The cool breeze on the ride, quite frankly, was soothing.</p><p>We stopped for lunch at a roadside dhaba. We sat inside the tiny huts at the dhaba; the waiter was looking at the menu for the first time, like us. We ordered 2 non-veg Assamese thalis &amp; Chinese items. The food wasn’t bad but wasn’t great either. But compared to all the food we had on the trip, this would definitely stay at the bottom. Even Prasidh Krishna wasn’t helping.</p><p>After the stop, we headed to Elephant Falls. Beautiful, clean place.</p><p>The Khasi name translates to “three-step waterfall”, because the water drops in three distinct tiers. The British renamed it after a rock near the falls that looked like an elephant’s head. The rock was destroyed in the 1897 earthquake, so the falls are now named after an elephant nobody can see.</p><p>We had serious doubts about our trekking capabilities when we panted, huffed &amp; puffed on the steep 150–200 m staircase around the falls. Ansh bought a poncho. TIL (Today I learned) what’s a poncho bc</p><p>Sushil had taken a sleeping pill (Ambien?) &amp; was sleeping throughout the car ride. While we were fucking around, Sushil had quietly gone inside the car &amp; dozed off.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/960/1*CRkcVQrNr_kItWKRQ8xsYw.jpeg" /></figure><figure><img alt="" src="https://cdn-images-1.medium.com/max/960/1*1kAO4rQdiaAWJGsLjY5uQg.jpeg" /><figcaption>Can we please address the elephant in the room?</figcaption></figure><h3>Police Bazaar, then into the dark</h3><p>Rudra was done with his interview. The plan now was to meet up at Police Bazaar, huddle up &amp; head straight to the resort at Tyrna, on the outskirts of Cherrapunji.</p><p>Police Bazaar was a bustling area. Shillong’s equivalent of Mall Road or Church Street. Big commercial spots, winding roads, people roaming around, lil shops at every nook &amp; corner, old-school vibe! We saw an old single-screen cinema showcasing Khasi films. Similar to the likes of Maratha Mandir.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*8VsAJ4ZfIhAt6--H7OWiHA.jpeg" /><figcaption>Single screen old school Khasi theatre. Had an awesome vibe.</figcaption></figure><p>While roaming the winding bazaars selling everything from clothes, bras &amp; silicone implants to electronics, we sat down at a mithai bhandar.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*fy9i24h2RirQPbHfK4CeIA.jpeg" /><figcaption>I really hope the Paw is a spelling mistake and not what I think it is</figcaption></figure><p>We spent a couple of hours over there, gulping down lassi, ladoos &amp; lots of water. There were many niche shops selling Khasi fare.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/864/0*lGvJnu4rGQQp47_Q.jpg" /></figure><p>Like any other group, we teased &amp; flirted with each other.</p><p>After a long wait, Rudra finally arrived. We reunited &amp; headed on our way to the resort in Tyrna.</p><p>One thing about Rudra is that he is a treasure trove of stories. On the dark, undulating, oft. scary roads, he spilled all the campus gossip, baddy team drama &amp; senior-junior interactions, and kept us entertained.</p><p>The road leading to the resort was so remote, narrow &amp; hilly that I had doubts. Forget the resort. Was there even a human settlement in Tyrna? How do these people live here? How do they travel? The network was wonky; how do they communicate?</p><p>Ansh, however, kept reassuring us that the resort would be good.</p><p>“Smoky Falls Valley Resort,” the signboard at the turn read.</p><p>FINALLY!!!<br>We had finally arrived!</p><p>We quickly settled into our rooms. After all the thulping &amp; gulping, and also to be ready for the trek the next day, I chose to eat something light for dinner. Chicken sandwich &amp; boiled veggies. The rest ate chicken curry &amp; roti. The food was ready, as we had pre-ordered it.</p><p>Interestingly, the tap water was also their drinking water. Such a European thing; in India, it was unheard of. I didn’t find anything odd about it when I drank it, &amp; also I was damn thirsty.</p><p>It was 10 pm already.</p><p>After dinner, we took a bath (separately) &amp; changed into our nightwear (separately). Sushil excitedly asked me to host a quiz. Rudra was flexing his credentials as a member of the BITS Goa Quiz Club to Harshit.</p><p>So I straightaway hosted a Sports Quiz, a three-way competition between Sushil, Harshit &amp; Rudra. It was very competitive, contrary to my expectations.</p><p>Ansh did all the heavy lifting of pre-ordering lunch, getting eggs &amp; apples for breakfast &amp; clearing all dues. He got his much-deserved shower while we quizzed. We discussed our plan for tomorrow &amp; retired for the night.</p><p>One incident that I found funny at this point was that Sushil went to our room to talk and locked the door for a very long time, while Ansh &amp; Rudra had slept in their room. I asked Harshit to ask Sushil to vacate the room so that we could sleep. I’m not sure what the relationship between Harshit &amp; Sushil is. Harshit admires him &amp; always wants to get in Sushil’s good books. He has already placed him up on a pedestal! Harshit is a damn genius, but he sometimes becomes a Sushil andhbhakt.</p><p>So when I asked him to politely ask Sushil to open the door, he hesitated! Lol.</p><p>Honestly, I don’t care if anyone else is present in the room if I call home. But Sushil is extremely conscious that way.</p><p>Not on my watch.</p><p>I asked him to open the door so that we could rightfully claim our beds to sleep. Sushil came out &amp; apologized in his trademark style. I apologized to him for interrupting his lovey-dovey call, but fuck it, we were hella sleepy.</p><p>Since we were leaving the next day, immediately after the trek, I packed my bags &amp; readied a backpack for the trek. I went around for a few minutes to enjoy &amp; appreciate the surroundings.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*tftsgKZ9lf7zAgn8uZ7Wuw.jpeg" /></figure><p>I was inside a cloud. <br>My head in the clouds.<br>Literally. Cloud Nine, with nobody else in it.</p><p>Maybe this is what bliss feels like. Absolutely amazing feeling.</p><p><em>Coming up next, Day 2: The trek &amp; how we BARELY survived it.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7f3a066198eb" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The time I met Dr. APJ Kalam]]></title>
            <link>https://medium.com/@topper1728/the-time-i-met-dr-apj-kalam-51c15d6147da?source=rss-78ce477380be------2</link>
            <guid isPermaLink="false">https://medium.com/p/51c15d6147da</guid>
            <category><![CDATA[prayag]]></category>
            <category><![CDATA[hyderabad]]></category>
            <category><![CDATA[stories]]></category>
            <category><![CDATA[abdul-kalam]]></category>
            <category><![CDATA[students]]></category>
            <dc:creator><![CDATA[Prayag Mohanty]]></dc:creator>
            <pubDate>Thu, 23 Jul 2026 22:55:44 GMT</pubDate>
            <atom:updated>2026-07-23T23:03:42.232Z</atom:updated>
            <content:encoded><![CDATA[<p><em>To be called a smart boy by THE Missile Man of India!! Unforgettable.</em></p><figure><img alt="" src="https://cdn-images-1.medium.com/max/767/1*rXzvPksxgq87zlJTQscqTA.png" /></figure><p>Back in 2014, former President of India Dr. A.P.J. Abdul Kalam visited my school, Little Flower High School, in Hyderabad to interact with the students.</p><p>Even though Pranab Mukherjee was the President, people still loved and considered Dr. Kalam their Prez. He was known to travel around the country, visit institutions and famously inspired young minds.</p><p>I was in 6th. Life was good.</p><p>As news of Dr. Kalam visiting our school reached our ears, our teachers were overjoyed &amp; went into overdrive with preparations. It was definitely a matter of honour for our school to host an eminent personality, the former President of India himself.</p><p>As per the event schedule, there was a QnA session where students could ask any questions to Dr.Kalam. However, the catch was that these questions had to be vetted in advance by our English teachers. Which meant students were to be filtered based on their questions.</p><p>What’s fascinating is that we had 2 selection rounds for this. The first round filtered out good questions from all the submissions. The second round had students read out the question in front of the teachers.</p><p>I don’t know what question I had submitted, but all I remember is that I was rejected after the 2nd round.</p><p>Is there any sense to this? <br>I mean, was I rejected for not pronouncing words well ?? <br>Idk. I was pretty dejected. I prayed for a miracle to happen.</p><h3>D-Day</h3><p>The day arrived. The school was decorated wonderfully. The energy was palpable. Kids from other schools also joined in.</p><p>This was one of the few occasions where there were chairs for us. WOAH. I sat somewhere near the front of the stage with my classmates. The kids who were selected to ask questions were asked to sit on the chairs along the carpet edge. I was lowkey jealous of them.</p><p>However, Dr. Kalam’s arrival was delayed by 30 min.</p><p>And then an hour…</p><p>2 hours….</p><p>But after a loooong wait, he finally arrived!</p><p>A bunch of bodyguards were around him, along with our school’s NCC cadets who marched ahead of him. He was much shorter than I expected. Screw that, he was of my height.</p><p>He quickly took the stage and began his speech.</p><p>During this speech, he asked a couple of questions about famous people and how they are remembered.</p><p>When we look at the bulb, Edison’s the name we remember.<br>Telephone. Alexander Graham Bell.</p><p>I happened to be the only one to answer his question,</p><blockquote>“ Who is the only woman to win 2 Nobel Prizes? “</blockquote><p>Being the competitive quizzer that I am, I&#39;m shouted with all my energy.</p><blockquote>“MADAM MARIE CURIEE!!!”</blockquote><p>He looked for the tiny voice among the crowd. He saw me &amp; acknowledged my answer.</p><blockquote>Very good fellow! Which class are you in?</blockquote><blockquote>6th</blockquote><p>Good answer.</p><p>Being called &quot;smart&quot; by THE missile man of India was the last thing I’d have thought would happen to me. Such an invaluable memory,</p><p>Damn, I was the happiest that day.</p><blockquote>What’s funny is, since the event was delayed, the QnA session was canceled :) Guess who had the last laugh ;)</blockquote><p>Though he never knew me (nor did I have the fortune to meet him later), he left such a deep impact on my life that I’ll never forget.</p><p>That incident continues to inspire to this day. <br>I realized words can truly inspire you. It tends to be more impactful if he is a visionary, someone who believed in the future of India.</p><h4><em>Still don’t believe me?</em></h4><p><strong><em>Watch the following video from 2:36</em></strong></p><iframe src="https://cdn.embedly.com/widgets/media.html?src=https%3A%2F%2Fwww.youtube.com%2Fembed%2FrwoVQF1_PAg%3Ffeature%3Doembed&amp;display_name=YouTube&amp;url=https%3A%2F%2Fwww.youtube.com%2Fwatch%3Fv%3DrwoVQF1_PAg&amp;image=https%3A%2F%2Fi.ytimg.com%2Fvi%2FrwoVQF1_PAg%2Fhqdefault.jpg&amp;type=text%2Fhtml&amp;schema=youtube" width="640" height="480" frameborder="0" scrolling="no"><a href="https://medium.com/media/7a0d634a99bdbcd33711b082dda37f83/href">https://medium.com/media/7a0d634a99bdbcd33711b082dda37f83/href</a></iframe><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=51c15d6147da" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Notes from 2023]]></title>
            <link>https://medium.com/@topper1728/notes-from-2023-7b53b5be0552?source=rss-78ce477380be------2</link>
            <guid isPermaLink="false">https://medium.com/p/7b53b5be0552</guid>
            <category><![CDATA[pilani]]></category>
            <category><![CDATA[notes]]></category>
            <category><![CDATA[bit]]></category>
            <category><![CDATA[arena]]></category>
            <category><![CDATA[hyderabad]]></category>
            <dc:creator><![CDATA[Prayag Mohanty]]></dc:creator>
            <pubDate>Wed, 01 Oct 2025 19:00:14 GMT</pubDate>
            <atom:updated>2025-10-01T19:00:14.565Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Found a draft from 2023, journaling my thoughts on my form and experience during Arena’23, the sports fest of BITS Hyderabad</em></p><p>I had practised badminton for days (2 weeks to be precise) 9:30 to 1130 pm every night. Strokes never went in place , it seemed as if I had lost my touch or forgotten how to play. Was it my racquet? Was it my bad muscle memory? Or both?</p><p>I was restlessly grinding to get back in form. Nights after practise was having dinner at ANC and heading to Malik’s room and working.</p><p>Finally the D day arrived, 1st Feb the day we leave. The bus arrival time was 5pm. I packed everything except for my passport sized photo (which Keshwani helped me take a pic of) I met my buddies and left for the bus. I waited in the flag lawns, got my passport pic printed from Pragati, saw the other half of the team leave for the airport. <br>Finally after 2 hours the buses arrived.<br>We were alloted the one with the UFC team and snooker team. At the gate, we waited for another hour for ID check and then we left. Inside the bus, me Anchal amogh and anurag played ludo chopsticks and slept. Anchal had made a delicacy from fafda and mayo and it was a good snack. The bus stopped for dinner at a dhaba, where there was bhaji aloo sabzi chapathi and good dal. Saw the draws. Slept for a good while. Stopped for breakfast. Had idli.</p><p>We finally reached BITS Hyd and whatta campus! The fountain wasn’t working but was a sight to behold. We rushed in through registrations got the useless passes and headed straight to acco at Viswakarma hostel common room aka VK.</p><p>Interconnected hostels- confusing but spacious. Went inside the common room took my spot, the tennis and chess team were with us. Maliwal, Hreetik, Sandesh and chinmay were there. Got one near partner gaurang.</p><p>Got ready for the match. Went to Sac. Sushil Ansh Tanmay Vardhan smashed SRM straight sets.</p><p>We played the 1st match against Bhavans. Gaurang played exceptionally well, went to 3 sets, but lost. Anurag played well but too many mistakes. Smashes into the net.</p><p>I played the doubles, couldn’t get the strokes initially but somehow played. Gaurang played good smashes. Won the doubles. But 3rd singles was a downer. Harshit played well but couldn’t get past through. Bhavans was relatively easier imo.</p><p>We had the girls team event which no surprises we lost. While team A was playing against Loyola ( Vignesh was trashing Sushil, Ansh played a 3 setter with Shashank)</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=7b53b5be0552" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[A New Paradigm for Tunable Filters That Leverages the Miller Effect]]></title>
            <link>https://medium.com/@topper1728/a-new-paradigm-for-tunable-filters-that-leverages-the-miller-effect-68ec71fd0e0b?source=rss-78ce477380be------2</link>
            <guid isPermaLink="false">https://medium.com/p/68ec71fd0e0b</guid>
            <dc:creator><![CDATA[Prayag Mohanty]]></dc:creator>
            <pubDate>Mon, 18 Aug 2025 19:40:01 GMT</pubDate>
            <atom:updated>2025-08-18T19:40:01.271Z</atom:updated>
            <content:encoded><![CDATA[<p><em>Dr. Naveen introduces a novel approach to designing continuous-time tunable filters using the Miller effect, eliminating the need for tunable impedances. He presents practical applications including true-time delay circuits, pulse shaping systems, and voltage-controlled oscillators, along with insights into high-output impedance OTA design in scaled technologies.</em></p><p>Imagine trying to catch a whisper across a crowded room. Now, imagine if you could somehow “stretch” that brief, faint sound into a longer, clearer sentence, making it perfectly understandable. This isn’t science fiction; it’s a powerful concept at the heart of modern electronics, particularly in how we process lightning-fast signals. Researchers like Naveen K. at IIT Dharwad are pioneering the next generation of circuits that can precisely manipulate the “timing” of electronic signals, opening up incredible possibilities from seeing through walls to making our devices smarter and more efficient.</p><p>At its core, this work delves into <strong>broadband filters</strong>, circuits designed to manage signals across a wide range of frequencies. But what if you could not only filter signals but also control their very “speed” or “delay” with pinpoint accuracy? This ability is crucial for handling fleeting, high-speed electronic pulses, turning them into forms that our everyday digital brains can easily digest.</p><h3>The Art of Signal Shaping: Expanding and Compressing Time</h3><p>Think of a fleeting electronic pulse, perhaps one lasting only a fraction of a nanosecond. Such pulses carry a lot of information in a very short time, requiring incredibly fast and expensive analog-to-digital converters (ADCs) to capture them. But what if we could “stretch” this narrow pulse into a broader, longer one without losing its information? This is where <strong>variable-delay filters</strong> come into play, serving as an <strong>Expansion (Compression) Circuit</strong>.</p><p>The fundamental building block for this “time-stretching” or “time-squeezing” magic is often a Low Pass Filter (LPF). The heart of its variability lies in a parameter denoted as <strong>α (alpha)</strong>. The relationship between the input (vin) and output (vout) of this circuit is described by the transfer function H(s) = (1 + (1+α)sRC) / (1 + s2RC), where R and C are resistance and capacitance values, and ‘s’ is the complex frequency variable.</p><p><strong>For compression</strong>, the parameter <strong>α is increased linearly</strong>. Conversely, <strong>for expansion</strong>, α is <strong>decreased linearly</strong>. This clever manipulation of α allows the circuit to dynamically adjust how much the signal is delayed and how its frequency components are handled, effectively stretching or compressing the signal’s duration. The presentation shows how varying α alters both the gain and delay characteristics across different frequencies, demonstrating the dynamic control over signal shape.</p><p>IIT Dharwad’s research has successfully implemented such a system, showcasing a <strong>15-stage Expansion Circuit</strong>. Designed using both UMC 180-nm and TSMC 65-nm technologies, post-layout simulations reveal impressive results. For an input pulse with a half-power pulse width of just 200 picoseconds (that’s 0.0000000002 seconds!), their circuit achieved a <strong>stretch factor of 1.25</strong> with a minimal Root Mean Square Error (RMSE) of 1.5 mV. This means a very short, intense signal can be broadened, making it easier for subsequent, slower processing stages to handle.</p><h3>Real-World Vision: Where Variable Delay Shines</h3><p>Why is this ability to stretch or compress signals so groundbreaking? It’s all about making complex, high-speed applications feasible and more accessible. The <strong>applications of variable-delay filters</strong> are diverse and impactful, particularly in scenarios involving narrow, sporadic pulses:</p><ul><li><strong>Ground Penetrating Radar (GPR):</strong> Imagine surveying underground to find pipes, cables, or even archaeological artifacts. GPR systems send out short electromagnetic pulses and analyze the reflected “late waves”. Traditionally, processing these incredibly fast echoes requires <strong>high-speed Analog-to-Digital Converters (ADCs)</strong>, which are expensive and power-hungry. By using pulse expansion, these fast echoes can be stretched into broader signals, allowing them to be sampled effectively by <strong>lower-speed, more affordable ADCs</strong>. This makes GPR more practical for a wider range of applications.</li><li><strong>Through-Wall Imaging and Radar:</strong> This technology has transformative potential in security, search and rescue, and even autonomous navigation. Imagine detecting movement or objects behind obstacles without physical contact. Variable-delay filters are key to accurately processing the subtle reflections from such environments.</li><li><strong>Ultra-Wideband (UWB) Technology:</strong> UWB systems transmit data across a very wide frequency spectrum using very short pulses. Applications include high-precision indoor positioning, real-time tracking, and short-range, high-bandwidth communication. The ability to precisely control pulse shape and timing in UWB is crucial for its performance and power efficiency.</li></ul><h3>Engineering the Delay: From Concepts to Circuits</h3><p>Achieving precise, controllable delay isn’t a simple feat. The research builds upon and advances existing techniques in variable-delay filters. Previous methods often involved varying bandwidth or using programmable transconductance (Gm) elements. These reported techniques aim to increase filter delay by reducing bandwidth, often using constant-capacitance © frequency scaling with programmable Gm. The challenge lies in ensuring that higher-order filters and Gms can be switched instantaneously and precisely.</p><p>The team’s work also explores how differential ring oscillator circuits can be leveraged for variable delay. These circuits can achieve delay variation by using <strong>varactors</strong> (voltage-controlled capacitors) or <strong>switched capacitors</strong> to change the delay of an LPF stage. Another method involves deriving <strong>variable resistance (R)</strong> from voltage-controlled PMOS transistors, integrating variable-delay LPFs directly into ring-VCOs (Voltage-Controlled Oscillators).</p><p>A significant contribution from this work is a <strong>3-Stage VCO with a Proposed Delay Cell</strong>. This innovative design aims to tune the oscillator frequency by changing capacitance through the <strong>Miller effect</strong>. Crucially, this approach <strong>does not require high-Q varactors</strong>, which simplifies design and can lead to more stable and reliable circuits while maintaining a constant oscillation amplitude. This is a key technical advancement, making the designs more robust and practical.</p><h3>Precision in Timing: The True Time Delay Advantage</h3><p>While variable-delay filters offer great flexibility, some applications demand a highly linear and precise control over signal delay, regardless of frequency. This is where <strong>True Time Delay (TTD) circuits</strong> become essential. TTD circuits aim to provide a delay that is constant across a broad range of frequencies, critical for maintaining signal integrity, especially for wideband applications.</p><p>The presented work introduces <strong>TTD with Delay Correction</strong>, a sophisticated approach to ensure precise timing. The concept involves meticulously determining parameters like β (beta) to ensure that the sum of the derivatives of delay with respect to frequency (dτ_α/dω + dτ_β/dω) equals zero, leading to a constant delay. For instance, specific settings like α = -2 and β = 1.6 can yield a constant delay up to a certain frequency (ω_g).</p><p>The researchers have successfully designed and simulated TTD circuits, demonstrating their practical implementation:</p><ul><li>A <strong>15-stage TTD circuit</strong> designed in <strong>TSMC 65-nm technology</strong> shows excellent post-layout simulation results, with consistent gain and precisely controlled delay across different settings and frequencies. The complexity of such a circuit is evident in its detailed block diagram and physical layout.</li><li>Similarly, a <strong>10-stage TTD circuit</strong> designed in <strong>UMC-180 nm technology</strong> also demonstrates robust post-layout simulation results, showcasing the scalability and adaptability of their design across different fabrication processes. These results are crucial in validating the theoretical concepts and practical viability of their TTD solutions.</li></ul><h3>The Numbers Don’t Lie: Benchmarking Performance</h3><p>For any cutting-edge electronics research, performance metrics are paramount. The presentation includes a comprehensive <strong>Results and Comparisons</strong> table, pitting “This work” against several reported variable-delay filter designs from major conferences and journals (MTTS-2019, SSC-2017, ISSCC-2019, MWCL-2007). This rigorous comparison uses several key parameters:</p><ul><li><strong>Technology (nm):</strong> The fabrication process node used (e.g., 65 nm, 180 nm).</li><li><strong>Maximum Delay (ps):</strong> The longest achievable delay.</li><li><strong>Bandwidth (GHz):</strong> The operating frequency range.</li><li><strong>Delay Range (ps):</strong> The total span of controllable delay.</li><li><strong>Power (mW):</strong> The power consumed by the circuit.</li><li><strong>Figures of Merit (FOMs):</strong> Two custom figures of merit are used to evaluate overall efficiency and performance:</li><li><strong>FOM-1 = 20log(Power / (delay range × bandwidth))</strong> — This metric suggests a preference for lower power consumption relative to the circuit’s delay range and bandwidth capabilities.</li><li><strong>FOM-2 = FOM-1 + 20log((2 × delay variation) / delay range) + 2 × gain variation</strong> — This second FOM adds a penalty for variations in delay and gain, emphasizing stability and precision.</li></ul><p>While specific numbers for “This work” vary slightly depending on the “LPF with corner” or “LPF without corner” architecture, the comparison highlights their competitive or superior performance in critical areas like <strong>power consumption, delay range, and overall FOMs</strong>, demonstrating the efficiency and precision of their proposed designs compared to the state-of-the-art.</p><h3>Beyond Filters: A Glimpse into Tomorrow’s Electronics</h3><p>The work by Naveen K. and his team at IIT Dharwad represents a significant leap in our ability to precisely control signals, offering a powerful toolkit for a range of demanding applications. The implications extend beyond just advanced radar and imaging.</p><p>The research group is also actively engaged in other fascinating areas, showcasing the breadth of their expertise:</p><ul><li><strong>Instrumentation Circuits:</strong> They are developing sophisticated circuits for MEMS (Micro-Electro-Mechanical Systems) and biopotential signals. This includes transistor-based MEMS sensors, such as Suspended Gate FETs, and their necessary interface circuits. This demanding work involves everything from device design and fabrication to circuit design, often in collaboration with institutions like IIST (with Prof. Seena V). These circuits require specialized components like analog amplifiers, high voltage charge pumps for actuator drivers, and efficient data converter circuits.</li><li><strong>Hearing Aid Design:</strong> In collaboration with Prof. Niyta Tiwari at IITBbs, the team is undertaking a ground-up design of hearing aids. This holistic approach encompasses signal processing, the development of custom Application-Specific Integrated Circuits (ASICs) for the electronic components, and comprehensive product design.</li></ul><p>This multi-faceted research portfolio underscores a dedication to not only advancing fundamental semiconductor and circuit design but also translating these innovations into tangible solutions for real-world challenges, from enhancing our sensory capabilities to improving quality of life. As electronics continue to shrink and demand higher performance, the precise control of signal timing, as demonstrated by their variable-delay and TTD circuits, will be an increasingly vital component in shaping the future of smart, connected, and intelligent systems.</p><p><strong>About the speaker:<br></strong><em>Dr. Naveen Kadiyanti holds a dual MTech + PhD degree from IIT Bombay’s Department of Electrical Engineering. He is an assistant professor at IIT Dharwad, and his research interests include mixed-signal circuits and instrumentation. Before IIT Dharwad, he was a postdoctoral fellow at IIT Madras and a visiting faculty member at IIST Thiruvananthapuram.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=68ec71fd0e0b" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Towards Multi-Chip Distributed Accelerators for Probabilistic Algorithms]]></title>
            <link>https://medium.com/@topper1728/towards-multi-chip-distributed-accelerators-for-probabilistic-algorithms-25d8627753a4?source=rss-78ce477380be------2</link>
            <guid isPermaLink="false">https://medium.com/p/25d8627753a4</guid>
            <category><![CDATA[ai]]></category>
            <category><![CDATA[distributed-systems]]></category>
            <category><![CDATA[accelerator]]></category>
            <dc:creator><![CDATA[Prayag Mohanty]]></dc:creator>
            <pubDate>Sun, 17 Aug 2025 11:27:06 GMT</pubDate>
            <atom:updated>2025-08-17T11:30:53.284Z</atom:updated>
            <content:encoded><![CDATA[<p>Imagine a world where the very foundation of our digital prowess, the silicon chip, is reaching its fundamental limits. For decades, we’ve relied on Moore’s Law to double our computational power, leading to the AI revolution we see today. But what happens when that predictable march of progress slows to a crawl? What if the most challenging problems, from designing new drugs to optimizing supply chains, require a different kind of thinking — one that embraces uncertainty rather than shying away from it?</p><p>This isn’t a distant sci-fi scenario; it’s the very real challenge facing computational science today. As transistor miniaturization struggles for every last nanometer, the improvements in computational speed and energy efficiency are hitting a wall. Our classical accelerators, like the ubiquitous GPUs and powerful TPUs, are masters of deterministic algorithms — tasks where every step is precise and predictable. But many of the cutting-edge problems in AI, combinatorial optimization, and probabilistic sampling don’t work that way. They involve <strong>probabilistic tasks</strong> like Monte Carlo and Markov Chain Monte Carlo (MCMC) algorithms, which are crucial for exploring vast solution spaces. On current hardware, these probabilistic algorithms are remarkably inefficient, burdened by their serial nature and the high-power consumption needed to generate random numbers. Even quantum computing, with its immense promise, grapples with significant hurdles such as cryogenic cooling requirements and high error rates, making it impractical for many immediate applications.</p><p>So, what’s the path forward when the traditional road is getting bumpy? This was the electrifying question explored by <strong>Tathagata Srimani, an Assistant Professor from Carnegie Mellon University (CMU)</strong>, during his invited lecture at the GG conference room on August 1, 2025, titled “<strong>Towards Multi-Chip Distributed Accelerators for Probabilistic Algorithms</strong>”. His talk illuminated a revolutionary direction: <strong>probabilistic computers built from probabilistic bits, or p-bits</strong>.</p><h3>The Rise of the Probabilistic Bit (p-bit): Embracing Randomness</h3><p>At the heart of this new paradigm are p-bits. Forget the rigid 0s and 1s of conventional computing for a moment. <strong>P-bits are classical bits that don’t just sit in one state; they fluctuate probabilistically between 0 and 1, all at room temperature</strong>. Think of them as tiny, highly controllable dice constantly rolling, providing the inherent randomness that probabilistic algorithms crave, without the energy cost of <em>generating</em> it externally.</p><p>Currently, various p-bit implementations exist, from <strong>analog systems</strong> like stochastic Magnetic Tunnel Junctions (MTJs) and oscillators, to <strong>FPGA-based demonstrations</strong>. While promising, these early versions have their own sets of limitations. Analog systems often face challenges with speed and scalability, while FPGA implementations, though flexible, can suffer from redundant logic that hinders both energy efficiency and speed. Clearly, a more integrated and optimized approach is needed.</p><h3>A Holistic Revolution: Application-Architecture-Technology Co-Design</h3><p>Professor Srimani’s lecture outlined a compelling pathway to scale these probabilistic accelerators, emphasizing a <strong>holistic application-architecture-technology co-design</strong>. This isn’t just about tweaking one part of the system; it’s about innovating simultaneously across the entire stack, from the high-level algorithms down to the fundamental transistors.</p><h4>1. Application-Level Innovations: Thinking in Graphs</h4><p>At the application level, a key innovation lies in <strong>graph-aware update scheduling</strong>. This technique is designed to preserve “constant-frequency sweeps” as the number of p-bits in a system increases. Why is this important? Many complex optimization problems can be mapped onto graphs, where nodes represent variables and edges represent their relationships. Efficiently navigating these graphs is paramount for solving problems like those tackled by probabilistic algorithms.</p><p>The nature of these graphs dramatically impacts how efficiently they can be processed by hardware. Consider the different types of graphs illustrated in the talk:</p><ul><li><strong>Sparse Graphs (e.g., Lattice, King’s):</strong> These graphs have relatively few connections between their nodes. They are <strong>hardware scalable</strong>, meaning they can be efficiently implemented on current architectures. However, they are limited to mapping “few problems”. Imagine a simple grid or a chessboard where connections are very localized.</li><li><strong>Resource-optimal Dense Graphs (e.g., Chimera, Pegasus):</strong> These represent a sweet spot. They are still <strong>hardware scalable</strong> and crucially, allow for <strong>“many problems (Graph Embedding)”</strong> to be mapped onto them. Think of them as more interconnected, specialized structures that balance complexity with hardware feasibility. These are the workhorses for probabilistic algorithms in AI and optimization.</li><li><strong>Densest Graphs (e.g., Custom, All-to-all):</strong> These are graphs where almost every node is connected to every other node. While “any problem can be mapped” onto them conceptually, they are unfortunately <strong>“not scalable”</strong> on current hardware due to the sheer number of connections required. Trying to build hardware for these would be like trying to build a city where every single building is connected directly to every other building — immensely complex and impractical.</li></ul><p>This understanding of graph properties is vital. By designing algorithms that intelligently map problems onto “resource-optimal dense graphs” and using graph-aware scheduling, researchers can unlock the true potential of probabilistic accelerators without being bottlenecked by the underlying hardware.</p><h4>2. Architecture-Level Breakthroughs: Tiling for Scale</h4><p>Moving down the stack, the architectural innovations are equally groundbreaking. Professor Srimani highlighted the development of <strong>CMOS p-bit prototypes that integrate tens of thousands of units</strong>. This is a crucial step towards practical systems, as it demonstrates the ability to pack a high density of these probabilistic elements onto a single chip.</p><p>But the real magic for scalability comes from how these prototypes are connected: they are <strong>tiled together via low-latency links into massive 30–50 k-spin clusters</strong>. Imagine connecting multiple powerful computing modules, each containing thousands of p-bits, with incredibly fast communication channels. This distributed multi-chip approach allows for computational power far beyond what a single chip could achieve, sidestepping the “Moore’s Law” limitations by leveraging parallelization and efficient inter-chip communication.</p><h4>3. Technology-Level Advances: Randomness at the Transistor</h4><p>Finally, at the foundational technology level, the innovation involves creating CMOS-based devices that inherently embed randomness. An example mentioned is <strong>perimeter-gated SPADs (Single-Photon Avalanche Diodes)</strong>. These devices are engineered to have <strong>bias-tunable randomness at the transistor level itself</strong>.</p><p>What does this mean? It means the random fluctuations that p-bits need aren’t being generated by external circuits or complex algorithms. Instead, the randomness is an intrinsic property, directly accessible and controllable at the very building blocks of the circuit. This deep integration of randomness at the transistor level has a profound impact: it can <strong>“further improve p-bit area efficiency for real-world optimization problems,”</strong> making these probabilistic accelerators even more compact and energy-efficient.</p><h3>The Road Ahead: A New Era of Computing</h3><p>The vision presented by Professor Srimani paints a compelling picture of a future where multi-chip distributed accelerators for probabilistic algorithms are not just a theoretical concept but a tangible reality. By addressing the limitations of Moore’s Law and the specific inefficiencies of probabilistic tasks on current hardware, this co-design approach offers a powerful alternative to both traditional deterministic computing and the challenges facing quantum computing.</p><p>This isn’t just about faster calculations; it’s about enabling solutions to previously intractable problems in AI, combinatorial optimization, and probabilistic sampling. As we continue to push the boundaries of what computers can do, these innovative probabilistic accelerators, with their ability to embrace uncertainty and explore vast possibilities, may well be the key to unlocking the next wave of technological breakthroughs. The journey towards truly intelligent and efficient AI systems just got a whole lot more exciting.</p><p><strong>About the speaker:<br></strong><em>Tathagata Srimani is an assistant professor at Carnegie Mellon University, ECE. Previously, he was a postdoctoral scholar in EE at Stanford University. He received the M.S. and the Ph.D. degrees in EECS from Massachusetts Institute of Technology in 2018 and 2022, respectively, and the B.Tech degree in E&amp;ECE from IIT Kharagpur in 2016. Tathagata’s research focuses on improving the energy-efficiency and throughput of computing systems hardware through advances in new technologies, heterogeneous integration, and technology–architecture codesign. His work has won the Best Paper Award at the 2023 IEEE Symposium on VLSI Technology and was highlighted in the U.S. President’s 2024 Nanotechnology Budget Report. He was a recipient of the MIT Presidential Fellowship in 2016 and the Morris Joseph Levin Award — best Masterworks (M.S. thesis) presentation at MIT in 2018.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=25d8627753a4" width="1" height="1" alt="">]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Scheduling for AoI: Distributed Protocols, Correlated Sources, and Non-Uniform Update Sizes]]></title>
            <link>https://medium.com/@topper1728/scheduling-for-aoi-distributed-protocols-correlated-sources-and-non-uniform-update-sizes-bc6cffcdee8d?source=rss-78ce477380be------2</link>
            <guid isPermaLink="false">https://medium.com/p/bc6cffcdee8d</guid>
            <category><![CDATA[distributed-protocol]]></category>
            <category><![CDATA[aoi]]></category>
            <category><![CDATA[correlated-sources]]></category>
            <category><![CDATA[scheduling]]></category>
            <category><![CDATA[acsm]]></category>
            <dc:creator><![CDATA[Prayag Mohanty]]></dc:creator>
            <pubDate>Sun, 17 Aug 2025 11:25:43 GMT</pubDate>
            <atom:updated>2025-08-17T11:25:43.016Z</atom:updated>
            <content:encoded><![CDATA[<p>Imagine a world where every piece of information you receive is instantly outdated. Your navigation system tells you to turn left, but the street was closed five minutes ago. Your smart home security alerts you to a potential intruder, but they’ve already left. In our increasingly connected world, where real-time decisions drive everything from autonomous vehicles to critical infrastructure monitoring, <strong>information freshness isn’t just a luxury — it’s a necessity</strong>. But how do we ensure that the vast streams of data flowing from countless sensors arrive at their destination not just quickly, but truly <em>fresh</em>? This is the core challenge tackled in a fascinating recent technical talk, delving into the intricate dance of wireless scheduling to keep our digital world up-to-date.</p><p>At the heart of this challenge lies a crucial metric: <strong>Age of Information (AoI)</strong>. Think of AoI like the “staleness” of a news update. It’s the time elapsed since the generation of the information that is currently available at the receiver. Minimizing this “staleness” is paramount.</p><h3>The Pulse of Information: Understanding the Standard Setup</h3><p>Consider a scenario with <strong>multiple sources (sensors) monitoring various processes, all relaying their updates to a central Base Station</strong>. This is “The Standard Setup”. The catch? There are <strong>interference and bandwidth constraints</strong>, meaning only a small subset of these sources can transmit at any given time. The primary goal is to <strong>design wireless scheduling policies that optimize monitoring performance by minimizing the average AoI</strong> across all sources. This complex optimization problem aims to find the best scheduling policy (pi) to minimize the time-average weighted sum of AoI values.</p><h3>The CSMA Race: Making Wireless Smarter for Freshness</h3><p>How do these sources decide who gets to transmit and when? One common approach is <strong>Carrier Sense Multiple Access (CSMA)</strong>, a decentralized method where devices “listen” before transmitting.</p><h4>Idealized Frame-Based CSMA: The Timer Game</h4><p>In an <strong>Idealized Frame Based CSMA</strong> system, the process is akin to a rapid-fire contest. <strong>Each source generates a random timer, </strong><strong>Z_i(t)</strong>. The source that comes up with <strong>the shortest timer wins the race</strong>. Once a source wins, it <strong>occupies the channel for the entire frame</strong> and gets to transmit its packet. Typically, Z_i(t) is modeled as an exponential random variable with a fixed parameter. This means the likelihood of a source winning depends on its generated timer, creating a distributed way to decide access.</p><h4>Fresh-CSMA: Making Timers Smart with AoI</h4><p>The real innovation comes when we make these random timers smarter. The concept of <strong>“Replicating Max-Weight”</strong> introduces a key idea: <strong>choose random timers that are <em>dependent</em> on the current Age of Information (</strong><strong>A_i(t))</strong>. Specifically, Z_i(t) are now exponential random variables with a parameter alpha * A_i(t). This subtle change means that sources with <strong>older information (higher </strong><strong>A_i(t)) are more likely to generate a shorter timer and thus win the channel access race</strong>, prioritizing the delivery of stale data.</p><p>The scheduling decision, pi^Fresh-CSMA(t), is simply to pick the source i that minimizes Z_i(t). This elegant approach allows a <strong>distributed CSMA mechanism to replicate the decisions of a centralized Max-Weight policy</strong>. By setting the parameter alpha large enough, the Fresh-CSMA policy can match the Max-Weight decision with high probability. Crucially, this distributed method offers a <strong>performance guarantee</strong> over the entire time-horizon, achieving long-term time-average performance similar to, and within a factor of 2 of, the centralized max-weight policy. This is a significant breakthrough, showing how a distributed protocol can achieve near-optimal centralized performance.</p><h4>Stepping into Reality: Near-Realistic CSMA</h4><p>While idealized models are useful, real-world networks introduce complexities. In <strong>Near Realistic CSMA</strong>, issues like <strong>collisions</strong> arise if multiple sources generate the same shortest timer. When this happens, the channel might remain empty for a duration, leading to “back-off overhead”. Parameters B and beta are introduced to control the <strong>trade-off between collision probability and back-off overhead</strong>, allowing for tuning the protocol&#39;s behavior in more practical settings. The simplified CSMA/CA analysis typically makes strong assumptions about continuous time backoff timers, instant carrier sensing, and perfectly synchronized frame-based access, which are relaxed in a more realistic model.</p><h4>The Proof is in the Pudding: Simulation Results</h4><p>The practical efficacy of these approaches is demonstrated through <strong>simulation results</strong>. When plotting the “Normalized Weighted Sum Average Age” against the “Network size N,” we see compelling evidence. <strong>Fresh-CSMA and Near-Realistic CSMA both perform remarkably well</strong>, showing a lower average age compared to a basic “Max-Weight” policy, especially as the network size increases. While “Optimal Stationary Randomized” still sets the benchmark, the proposed CSMA variants offer a distributed and efficient way to achieve competitive performance.</p><h3>Beyond Independent Silos: Correlated Monitoring</h3><p>Traditional work on monitoring often assumes that each sensor tracks <strong>independent processes, with no correlation between them</strong>. However, in real-world applications, this is rarely the case. Imagine a network of environmental sensors where temperature readings in one area often correlate with those nearby, or a surveillance system where one camera’s view might provide partial information about events captured by another. This scenario introduces <strong>“Correlated Monitoring”</strong>.</p><p>In such a system, the problem becomes two-fold and <strong>tightly coupled</strong>:</p><ol><li><strong>Estimation:</strong> How to <strong>optimally reconstruct the state of all processes</strong> using past updates and the inherent correlation structure of the system. This often involves advanced techniques like the <strong>Kalman Filter</strong> for state estimation, especially for “oblivious scheduling policies” where estimation is separate from scheduling.</li><li><strong>Scheduling:</strong> How to <strong>choose which sensor gets to transmit at each time</strong> to minimize error.</li></ol><p>The overall goal is to <strong>minimize the time-average expected monitoring error</strong>, which is defined as the mean squared difference between the estimated and actual states. This monitoring error is deeply connected to AoI, as it can be <strong>bounded above and below by the AoI</strong> of the same process. This means keeping information fresh is directly linked to keeping our understanding of the system accurate.</p><p>This optimization problem transforms into a <strong>dynamic program</strong>, taking into account the system dynamics (where states evolve over time) and scheduling constraints. However, naive approaches like <strong>“Naive Max-Weight” policies</strong> face challenges here. Unlike simple queueing problems, the Lyapunov function (a common tool for stability analysis) isn’t directly suitable for AoI with heterogeneous update sizes. Why? Because the reduction in AoI from a transmission might happen in the distant future, and it can depend on the <em>stochastic outcome</em> of future scheduling decisions, making short-sighted optimizations ineffective.</p><h3>The Juggling Act: Large and Small Updates</h3><p>Another crucial real-world consideration is the <strong>heterogeneity of information updates</strong>. In many time-sensitive applications, the content isn’t uniform; it includes a <strong>mix of small information updates and large information updates</strong>. Think of location coordinates (small, frequent) versus streaming video (large, continuous), or LIDAR point scans mixed with control messages.</p><p>The network model reflects this: sources generate updates consisting of L_i packets, where L_i can vary. Scheduling then happens at the packet level, with each packet taking one time-slot to transmit. This introduces interesting dynamics for AoI evolution, where different update lengths lead to different AoI growth patterns.</p><p>This complexity raises an important question: <strong>Can we still achieve AoI minimization in this setting using Max-Weight style policies?</strong> The answer is a resounding <strong>Yes!</strong>. Even with this mix of update sizes, the underlying principles of prioritizing stale information can be effectively applied. Research in this area, such as “Optimizing AoI in Networks with Large and Small Updates,” continues to push the boundaries of what’s possible.</p><p>Intriguingly, when exploring different scheduling policies in a 2-source network with varying update lengths, it was observed that <strong>many policies discussed in the literature (e.g., Max-Age First) behave quite similarly to a “No-Switching” policy</strong>. This suggests that while optimizing for freshness, the complexity of constantly switching between sources might not always yield significant benefits in certain scenarios, highlighting the nuanced nature of optimal scheduling.</p><h3>Beyond the Horizon: The Future of Fresh Information</h3><p>The journey to consistently deliver fresh information in dynamic wireless environments is a complex, multi-faceted challenge. From the elegant simplicity of CSMA-based distributed scheduling to the intricate dance of estimation and scheduling in correlated networks, and the pragmatic considerations of handling diverse data types, researchers are continuously refining how our devices communicate. As our reliance on real-time data grows across every sector — from smart cities and industrial automation to healthcare and entertainment — these advancements in optimizing information age will be pivotal, ensuring that the insights we gain from our connected world are always timely, accurate, and truly fresh. The silent race for fresh data continues, driving innovation at the very core of our digital future.</p><p><strong>About the speaker:</strong><br><em>Vishrant Tripathi is an Assistant Professor in the ECE department at Purdue University, where he leads the Multi-Agent Intelligent Networks (MAIN) group. His research interests lie in the optimization of networks — with application areas including multi-agent robotics, networked control, federated learning, and edge computing. He obtained a Ph.D. in the EECS department at MIT, B.Tech from IIT Bombay and has also spent time working on data-center network optimization at Google.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=bc6cffcdee8d" width="1" height="1" alt="">]]></content:encoded>
        </item>
    </channel>
</rss>