Weekly Reading List: 29/11/2025
Things I Read This Week That Made Me Think
I’ve been thinking about the difference between a story and an excuse.
A story is what you tell when you want someone to understand how something came to be. An excuse is what you tell when you want them to stop asking. The tricky part is that they often sound identical. “I discovered it by accident.” “I had no choice.” “That’s just how it happened.” Sometimes true. Sometimes a way of skipping over the parts that would make the listener uncomfortable or implicate them.
This week’s reading kept pulling at that thread. The Penicillin Myth isn’t really about Fleming; it’s about how we flatten decades of methodical preparation into a single lucky afternoon because the flattened version is easier to teach and easier to believe. The coolgirl essay isn’t really about MTV VJs; it’s about how performance gets repackaged as authenticity every ten years, and how the people performing often can’t tell the difference anymore. The OpenAI piece isn’t really about Gemini versus GPT; it’s about what happens when the story (”scrappy lab disrupts the giants”) runs headfirst into the truth (”platform economics always win eventually”).
There’s a version of this list that’s about “narratives versus reality,” but that framing feels too clean. The narratives aren’t lies, exactly. They’re just… incomplete in ways that serve someone. Usually the person telling them. Sometimes the person listening, who’d rather not know how the sausage gets made.
The question I kept returning to: what would it cost to tell the full version? And who’s been paying to keep the short version in circulation?
Let’s go.
1. How Not to Get Away With Murder: The Stranger-Than-Fiction Story of the Stoney Creek Killing Toronto Life — Sarah Treleaven
Lucy Li wanted to be a TikTok star. Oliver Karafa wanted to be a millionaire before 30. When a friend (and creditor) got in the way of their plans, they cooked up a murder scheme so incompetent that one of their lawyers literally used stupidity as a defence.
The piece walks through: Karafa’s prior conviction for dangerous driving, his prison stint, his post-release delusions of grandeur; Li’s Kardashian aesthetics and reality-TV aspirations; the Google Maps scouting of a secluded warehouse; the SIM-card swap to create a “digital alibi”; the plan to dispose of two cars; and the moment it all fell apart because one of their victims survived, crawled to the road, and flagged down help.
They fled to Europe. They were extradited. They were convicted of first-degree murder and attempted murder.
What I got out of it: A reminder that ambition without competence means nothing, and that the most dangerous people are often the ones who believe their own delusions. Also, a genuinely gripping true-crime narrative that earns its length.
2. Made for the Moment Hipcityreg
A framework essay on “The Moment”: when you command clear attention as the centre of the most important environment. The Moment isn’t necessarily positive; you can be The Moment because you’re fighting for your life after a fall from grace, or because you’ve been ascendant for a while.
Key claims:
The Moment can’t be faked. Marketing is not The Moment; it’s laying narrative toward a potential Moment.
The Moment demands previous suffering and obscurity. The Hero’s Journey arc is real.
Nepo babies struggle because the crowd has already condensed their attempt at Moment-ness down to their parents.
Understanding what is “built for The Moment” is a critical framework for tech and venture.
What I got out of it: A useful lens for thinking about narrative in tech. Sam Altman and OpenAI are The Moment right now, which is why even Gemini 3.0’s release gets referenced back to them. The framework also explains why so many “Head of Narrative” hires will fail: they’re hired to manufacture something that can only be adorned, not created.
3. The Coolgirl Was Never Real Poombatta — Medha
A memoir-essay about growing up in Trivandrum trying to decode the coolgirl: the MTV VJs, the school celebrities, the girls who seemed to have a secret manual for hair removal, short skirts, and being pulled up at assembly while also getting picked for Annual Day bouquets.
It traces a line from GTalk (an MTV show featuring two Delhi girls being deliciously unhinged in a fake apartment) to the Moment of Silence podcast (two Mumbai girls being strategically relatable while selling Pond’s Super Light Gel). The argument: the coolgirl code keeps mutating, but the basic mechanism (perform something aspirational, make it seem attainable, monetise the gap) hasn’t changed. What has changed is the mandate: in 2008, you had to be jarring; in 2025, you have to be a #girlsgirl.
What I got out of it: A meditation on relatability as product. The coolgirl was never real because she was always a performance, but the performance has shifted from “how dare she” to “she’s just like us.” The essay made me think about how much of what gets called “authenticity” is just a better-optimised sales funnel.
4. The Airport-Lounge Wars The New Yorker — Zach Helfand
A long, funny, slightly nauseating tour of the arms race in airport lounges. Capital One has a kitchen designed by José Andrés and a cart that wheels around delivering caviar cones. Delta One at JFK has a wellness area, shower suites, and a terrace with a retractable roof. Chase Sapphire has a hidden game room behind a photo booth.
The subtext: aspiration is now a subscription service. Lounges were once “cubes of cheese and second-rate wine” for business travellers. Now they’re lifestyle signifiers and the overcrowding has sparked its own backlash (NYT op-eds complaining that the riffraff are ruining the experience…LOL).
What I got out of it: A masterclass in writing about class without being preachy. The piece knows it’s absurd. It also knows that absurdity doesn’t make the comforts less real. I came away thinking about how “exclusive” spaces expand until they’re no longer exclusive, and then the truly rich just buy a new tier.
5. Between Plath and Didion: The Most Important Lesson I Learned in My Twenties Fairytales by Caroline
A personal essay about turning 30 and finally understanding the space between two quotes: Esther Greenwood’s paralysing fear (if I choose one fig, I lose all the rest) and Didion’s flippancy (nothing I do now will count).
The author’s twenties: brand consulting in New York (rotten fig), high school teaching in D.C. (rotten fig), then finally an MFA in fiction writing (the right fig). The advice: you have to pick a fig, but picking wrong doesn’t mean the tree disappears. Keep picking until you find the one that feeds you.
What I got out of it: A useful reframe for anyone stuck at the tree. The Plath posture is paralysing; the Didion posture is dismissive. The adult posture is: choices matter, but they’re not irrevocable. The goal isn’t to pick right the first time, but it IS to pick thoughtfully and be honest when something’s rotten.
6. The Cost of Becoming a Person hk’s newsletter
My own piece. It starts from the observation that “be yourself” is advice with hidden fees. The essay tries to map the actual ledger: what it costs to leave a role that was assigned to you, who bills you in guilt versus money versus silence, and why some people’s experiments are subsidised while others are punished.
I dig into:
The difference between acquisition cost (the dramatic moment you quit/leave/speak) and maintenance cost (the auto-debit of staying changed)
Why systems offer discounts for not growing
The surcharges that arrive with your body before you’ve ordered anything: gender, caste, class, geography
How self-help’s fantasy of “alignment” never mentions the fallout of walking into a room and announcing the old you is gone
What I got out of writing it: Clarity on something I’d been circling for years. The cost of becoming a person isn’t a one-time fee. It’s a subscription. The only leverage you have is choosing what you’re subscribing to.
7. The Conscious and Subconscious Mind: Everything You Need to Know Ixcarus
A synthesis of pop-psych and neuroscience on how the mind operates. The headline stat: your conscious mind processes ~50 bits per second; your total sensory input is ~11 million bits. Your conscious mind is handling 0.0004% of reality. The rest is subconscious pattern-matching.
Key concepts:
Reticular Activating System (RAS): Filters what gets to conscious attention. You program it with your focus, which is why you suddenly see your new car everywhere.
Emotional reactions come from subconscious programs, not conscious choice. Someone insults you, you feel anger instantly. That’s not a decision; it’s a script installed in childhood.
Reprogramming requires repetition and emotion. Conscious knowledge alone doesn’t overwrite subconscious patterns.
What I got out of it: A useful reminder that “you” (the conscious you) are mostly a passenger. The subconscious decides and the conscious mind rationalises. Not an argument for fatalism, but an argument for being more intentional about what you repeat, what you expose yourself to, and what environments you choose.
8. Mark’s 12 Principles of Watch Collecting Mark Cho
Mark Cho (of The Armoury) distils a lifetime of watch collecting into twelve principles. This is ostensibly about horology, but it’s really about attention, taste, and the discipline of knowing what you actually want versus what you’ve been told to want.
Some highlights:
“Rare watches are not always great, great watches are not always rare.”
“Expensive watches can be worth the cost. Cheap watches are not always worth the savings.”
“A perfect condition watch is priceless until you actually have to wear it.”
“All your watches want your time but you only have one wrist.”
“You can hold a watch forever but a watch cannot hold your attention forever.”
What I got out of it: The best collecting advice I’ve read in years, and it generalises far beyond watches. The core insight: appreciation scales with time spent wearing and understanding. The things that reveal themselves slowly are usually the ones worth keeping. And the hardest skill is learning to sell what you once loved, so it finds a home where it’s appreciated.
9. The Penicillin Myth Asimov Press — Kevin Blake
Everyone knows the story: Fleming goes on holiday, leaves a culture plate out, a mould floats in through the window, and eureka! penicillin. Except the story doesn’t hold up.
This essay digs into the scientific and historical problems: penicillin only works if it’s present before the bacteria grow, so the contamination couldn’t have happened after the staphylococci colonies were established. The window was rarely open. The first lab notebook entry isn’t until two months after the supposed discovery. And nobody has ever successfully replicated the “accidental” plate.
Two competing theories:
Hare’s theory (1970): A freak cold snap in London allowed the mould and staphylococci to be contaminated simultaneously, with the low temperature giving the mould time to produce penicillin before the bacteria grew.
Root-Bernstein’s theory (1989): Fleming wasn’t running a staphylococcus experiment at all. He was systematically searching for new lysozymes, and penicillin was a serendipitous byproduct of that methodical search.
What I got out of it: The myth of “pure accident” obscures the real lesson: Fleming was prepared to notice something unusual because he’d spent years building a framework for discovery. “Chance favours the prepared mind” is a cliché, but the essay earns it. Also: the logic of presentation rarely matches the logic of discovery. What looks like a lightning bolt in the paper was probably a slow accumulation of hunches.
10. ‘He Made Films, He Made Filmmakers’: How Did Jagat Murari Do It? Scroll — interview with Radha Chadha
Radha Chadha’s biography of her father, Jagat Murari, the man who built FTII into a world-class film school, is getting deserved attention. This interview unpacks how Murari created the curriculum, invited established filmmakers as guest lecturers, built the National Film Archives, and nurtured a generation of directors, actors, and cinematographers.
The key insight: Murari’s job was to make his students see beyond formulaic industry films. He did it by exposing them to world cinema (French, Soviet, Japanese) until the familiar became strange. “At first, this extensive exposure has an unsettling effect on students,” Murari noted. That was the point.
He was eventually pushed out by student strikes and bureaucratic failures. The institution he built nearly collapsed after his departure.
What I got out of it: A reminder that great institutions require both vision and management, and that the people who build them are often treated as expendable once the structure exists. Also: a useful case study in how “Indian cinema” was made, not born.
11. Active Learning Guides Automated Discovery of DNA Delivery via Electroporation for Non-Model Microbes bioRxiv preprint — Brumwell et al.
A technical paper from Cultivarium about building a robotic platform to discover electroporation protocols for microbes that don’t come with instruction manuals. Most microbes on Earth are “non-model”. We don’t know how to genetically engineer them because the basic step of getting DNA inside the cell is unsolved.
The team built a custom electroporator (most commercial ones are “black boxes” with preset parameters), combined it with a pooled plasmid library, and used active learning to iteratively improve protocols. They report the first electroporation protocols for six bacteria and significantly improved protocols for several others, including an 8.6-fold improvement for the industrial workhorse Cupriavidus necator.
What I got out of it: This is infrastructure for biological diversity. If you can’t get DNA into an organism, you can’t study it or engineer it. The paper is dry, but the implication is huge: a discovery platform that could unlock the “dark matter” of microbiology. It’s also a case study in how automation and machine learning can systematise what used to be bespoke, artisanal, and painfully slow.
12. The Structural Collapse: How Google’s Integrated Stack Is Dismantling the OpenAI Thesis Shanaka Anslem Perera
This piece argues that the AI narrative is inverting. OpenAI’s valuation (~$500B) rests on assumptions that are becoming harder to defend: that it can achieve Google-scale revenue while maintaining startup-level growth, that it can close the data/hardware/distribution gap, and that the Microsoft partnership is unconditional.
The structural advantages Google has:
Data: 4 billion users across Search, YouTube, Android, Gmail, Maps generating continuous behavioural data as a byproduct of services people already use.
Hardware: Custom TPUs, vertically integrated, reportedly 6x more efficient per watt than Nvidia H100s for transformer training.
Distribution: Gemini 3.0 rolled out directly into Google’s existing ecosystem. Zero adoption friction.
OpenAI, by contrast, rents Nvidia GPUs through Azure, has no consumer platform with billions of DAUs, and must convince users to visit a separate website. The leaked Altman memo (”rough vibes,” “wartime footing”) is read as a signal that even he no longer believes the growth trajectory is guaranteed.
What I got out of it: A cold shower for anyone still pricing OpenAI as if it’s inevitable. The key insight: model capability is commoditising, and the durable moats are platform economics (data, hardware, distribution). If AI value concentrates in platform companies rather than specialised labs, most venture-backed AI companies face margin compression and difficult exits.
13. The 400-Year Bubble Study: Inside Coatue’s AI Report The VC Corner
Coatue (Philippe Laffont’s $54B hedge fund) published a report arguing that AI is not a bubble but an early industrial revolution. They studied 30 bubbles over 400 years and concluded we’re still in the “displacement” phase, not “euphoria.”
Key slides:
AI stocks have outperformed the S&P 500 by 160%+ since ChatGPT’s launch.
The Nasdaq’s P/E peaked near 90x in 2000; today it’s ~28x.
IPO activity is muted compared to past bubbles (500+ IPOs/year in the dot-com era; fewer than 60 now).
Coatue assigns >66% probability to “AI Abundance” (productivity accelerates, inflation stays low) and <33% to “AI Reckoning” (bubble pops, recession follows).
What I got out of it: Depends on your priors. If you trust Coatue’s framing, this is reassurance that the current AI investment cycle is grounded in fundamentals. If you’re sceptical, it reads like a hedge fund talking its own book. The most interesting slide: the “infinite money loop” of vendor financing (OpenAI, Nvidia, Oracle circulating capital among themselves). Sustainable only if ROI keeps improving.
14. Sensex, Nifty Hit Record Highs: What It Means for Mutual Fund Investors Indian Express / various
The Sensex crossed 86,000 and Nifty topped 26,300 for the first time — the first record highs in 14 months. The rally was driven by strong FII inflows, optimism about US Fed rate cuts, and improved Q2 earnings.
Context:
Mutual fund SIP contributions hit ₹21,262 crore/month.
Total AUM for mutual funds is now ~₹46 lakh crore (doubled since 2017).
Retail participation is up, especially from Tier-2 and Tier-3 cities.
The shift from physical assets (gold, real estate) to financial assets (mutual funds) is accelerating.
What I got out of it: The standard “retail is finally getting equity exposure” narrative. Probably true, and probably a sign of financialisation deepening. The question I’m left with: what happens when the next correction arrives and all those new SIP investors discover that markets go down too?
15. Too Long: Your Prompt on ChatGPT Is Too Long Ruben
A practical guide to prompt length, backed by recent research. The core finding: 150–300 words is the sweet spot for moderate-complexity tasks; past 500 words, you’re likely confusing the model rather than clarifying.
Key concepts:
Lost-in-the-Middle: LLMs give less weight to information in the middle of long prompts.
Bloat ≠ clarity: Dumping entire documents rarely helps. It dilutes instructions and invites error.
Structure > length: Put context/persona at the beginning, passive data in the middle (fenced clearly), constraints at the end.
What I got out of it: Immediately useful for anyone who works with LLMs. The meta-point: your job is to reduce activation cost for the model, not to impress it with thoroughness. “Just because it can absorb a lot of text does not mean it does it well.”
There’s a moment in the Jagat Murari piece where his daughter describes finding his papers after he died. Three packed Godrej cupboards. Trunks of material. Preparatory notes for classes he taught decades ago. The infrastructure of a life’s work, sitting in storage, waiting for someone to make sense of it.
I think about that image a lot. Not because it’s romantic (it’s actually a little heartbreaking, all that labour collapsing into “stuff to sort through”) but because it’s honest. The things that matter don’t announce themselves. They accumulate. They sit in cupboards. They wait.
Most of the pieces this week are about that gap: between the moment something becomes visible and the years it took to become possible. Fleming’s “accident” was decades of looking at plates. Murari’s FTII was a curriculum built from scratch, guest lecturer by guest lecturer, while fighting budgets and bureaucrats. The coolgirl’s effortlessness is rehearsed. The airport lounge’s luxury is a credit card company’s customer acquisition cost. Even the murder in Stoney Creek, that disaster was the output of two people who’d spent years constructing a fantasy of themselves and finally ran out of room to keep it airborne.
I don’t know what to do with this except to say: the preparation is the thing. Not the reveal. Not the origin story you’ll tell later. The years of not-yet, the cupboards filling up, the slow accumulation of reps that nobody sees.
The story will come. Someone will write it, or you’ll tell it yourself, and it will be shorter and cleaner than what actually happened. That’s fine. Stories are supposed to be shorter.
Just don’t mistake the story for the work. And don’t let anyone else’s clean narrative make you feel like your own messy, unmarketable, cupboard-filling process is somehow wrong.
Until next week.