The Unread Pile

Edition No. 23 Saturday, 17 October 2026 21 new saves

The Lead Story

Turns Out the Future Does Have Flying Cars

Via @elonmusk, @Tesla, @MKBHD — and the timeline’s finest Z-axis comedians

Editorial illustration: a futuristic sports car flying through the air above a city skyline at dusk

Illustration: the reveal as stunt — Tesla’s Roadster takes the Z-axis, and the timeline does the rest of the marketing.

Elon Musk said it in seven words: “Turns out the future does have flying cars.” Tesla’s account followed with a 46-second video and two more: “Real life, baby.” Then came the specs, and this is where the launch went from stunt to spectacle. Marques Brownlee’s disbelief is now the canonical reaction: they got on stage and said a newly designed car weighing over 4,000 pounds can do flips, travel at more than 380 miles an hour — half the speed of sound — and will ship in 2028. “This is a prank, right?”

The design language, per Tesla’s own reveal post, has evolved from S3XY to Cybertruck, Cybercab and Optimus — and the Roadster takes its inspiration from the SR-71 and the F-35. Fighter jets, not cars. That framing is pure Tesla grammar: announce the impossible on stage, then let the timeline do the distribution. And the timeline obliged. Vijay Shekhar Sharma wondered aloud what the demo was even for, imagining the scenes in a traffic jam. Derek Nee noted that traffic is a bit heavy in the Z-axis today. Alexandr Wang asked the only honest engineering question: “wait who the hell was driving the tesla.” Priya thanked Musk for making Telugu-movie flying-car stunts “accessible to the normies.” Observirious is planning tomorrow’s commute accordingly. Deepinder Goyal managed only an emoji of disbelief. Gaut’s verdict: “narrator: he wasn’t joking.”

The counterweight to the laughter is Dilip Kumar’s four-word investment thesis: “Never bet against @elonmusk.” It’s the track-record argument — every previous “prank, right?” eventually shipped in some form, from reusable rockets to the Cybertruck’s entire existence. When the specs sound like a joke, the joke is doing the marketing: MKBHD’s incredulity reached more people than any spec sheet could.

4,000+

Pounds, newly designed car

380+

MPH — half the speed of sound

Flips

Yes, flips, on stage claim

2028

Ship date, per the reveal

“They really just got on stage and said this newly designed 4000+ pound car can do flips and go half the speed of sound (380+mph) and will ship in 2028. This is a prank, right?”— @MKBHD, on the Roadster reveal

Main learnings

Spectacle is the distribution. The reveal’s job isn’t to inform; it’s to make the timeline do the informing.
Disbelief is the funnel. “This is a prank, right?” is a more effective headline than any press release.
The track-record trade. “Never bet against Elon” is really a bet that past impossible-shipped predicts future impossible-shipped.
The Z-axis joke economy is the launch landing. When the memes write themselves, the product moment has already happened.

Practical takeaway — when a demo sounds like a prank, don’t ask whether the specs are real. Ask who’s laughing — and whether the laughter is the marketing working as designed.

Musk’s post · Tesla’s video · MKBHD’s reaction · Never bet against Elon

The Business

The ₹168-Crore Househelp

Via @pranavmukul — quoting ETtech’s numbers on Snabbit

Pranav Mukul surfaced the number of the week from ETtech: instant-househelp startup Snabbit posted a ₹168 crore loss on ₹12 crore of revenue in FY26. Read that ratio again — for every rupee of revenue, fourteen rupees of loss. Samidha Sharma’s original line said it plainly: “This is crazy cash burn!!!”

The mechanism is the collision of two playbooks. Quick-commerce trained a generation of founders that speed plus subsidies equals category creation. But househelp is the anti-quick-commerce: labour is the product, not the logistics layer around it. You can compress delivery with dark stores and routing; you cannot compress a human hour with a warehouse. The 14x burn multiple is what happens when the blitzscaling thesis meets the oldest unit-economics wall in Indian consumer internet — trust, retention, and wages that don’t scale down.

Snabbit FY26 — revenue vs loss (₹ crore)

Source: ETtech via @pranavmukul; figures for FY26.

Main learnings

Labour-heavy services don’t compress. The quick-commerce playbook assumes the costly bit is logistics; here the costly bit is the service itself.
Subsidy buys trials, not trust. Letting a stranger into your home is a trust purchase — discounts don’t move it the way they move a grocery order.
The burn multiple is a clock. 14x isn’t a growth investment; it’s a countdown to the round where the story has to change.

Practical takeaway — if your moat needs the customer to keep paying after the coupon ends, the coupon was the business. Price the service people rebook, not the one people try.

Pranav’s post with the ETtech numbers

The Culture

India’s VFX Coming-Out Party

Via @aviralbhat and @josephradhik — on the new Ramayana trailer

Editorial illustration: an epic mythological warrior scene rendered like a modern VFX film frame

Illustration: the new trailer’s promise — an Indian epic rendered at global VFX standard.

Aviral Bhatnagar’s verdict on the new Ramayana trailer is the kind of sentence an industry waits years to hear: the trailer is epic, and it’s showing India’s growing VFX prowess — visuals, score and composition all landing in unison. His prediction: it does very well globally, and opens to full houses in India within weeks. Joe’s review adds the craft note that matters: a visual spectacle is what this deserved to be — something the first teaser didn’t promise.

The mechanism underneath is unison. Indian epics have always had scale; what’s new is craft density — VFX, sound design and composition arriving together at a global standard, in the same frame. And the Ramayana is the perfect stress test for that claim: it’s a story every Indian viewer knows beat by beat, so the spectacle can’t hide behind novelty. It has to earn the awe against the audience’s own imagination, which has had thousands of years of head start.

“Ramayana’s new trailer is epic showing India’s growing VFX prowess. The visuals, score, composition are all landing extremely well in unison.”— @aviralbhat

Main learnings

Craft unison beats scale. One great department can’t carry a film; all of them landing together is the actual milestone.
Known stories are the hardest demo. When the audience already knows every beat, only craft can surprise them.
Domestic proof precedes global ambition. Full houses at home are the credential the global run will be sold on.

Practical takeaway — watch the credits, not the trailer. The VFX houses on this film are the industry’s next export — that’s where the compounding is.

Aviral’s review · Joe’s review

The Tech

A Bubble in a Bubble

Via @levelsio — replying to the “software is dead” provocation

Benja’s provocation set it off: “Software is dead. It’s now the era of personal software…” Levels’ reply is the sharper observation: he’d call this a bubble in a bubble — tech people somehow still think normies aren’t spinning up personal software too.

The mechanism is the cost collapse. When building software costs a weekend and twenty dollars instead of a dev team and six months, the constraint moves — from engineering to distribution and taste. Tech insiders keep score by who can build; meanwhile the normies are already building for themselves, no GitHub required. Two bubbles stacked inside each other: the AI investment bubble, and nested within it, the abundance bubble — infinite personal software, almost none of it defensible, all of it real.

What collapses when building costs ~nothing

Illustrative; the direction, not the numbers, is the point.

Main learnings

Abundance kills the builder’s moat. If anyone can build it, “we built it” stops being a pitch.
The new scarcities are distribution and taste. Getting it in front of people, and knowing what’s worth building, is where the value pools.
Insiders underestimate outsiders by default. “Normies aren’t doing this” has been wrong about every consumer technology wave.

Practical takeaway — if your product can be rebuilt in a weekend with a prompt, your moat was never the code. Find the part that can’t be vibe-coded: the data, the distribution, the trust.

Levels’ reply

Philosophy Grads Teach the Code to Think

Via @abhijitwt — on Anthropic’s new fellowships

Editorial illustration: a marble philosopher statue teaching a classroom of small robots

Illustration: the new classroom — reasoning is the curriculum, syntax is just the chalk.

Abhijit’s line of the week: CS grads spent four years learning to code; philosophy grads get four months teaching the code to think. The occasion: Anthropic opening applications for two new fellowships, including STEM Fellows — the frontier labs are now formally recruiting outside the traditional pipeline.

The mechanism is the skill-premium flip. Writing code was the scarce input; directing reasoning is the scarce input now — decomposing problems, spotting bad reasoning, knowing what “good” looks like before the machine produces it. Philosophy trains exactly that: argument structure, counterfactuals, definitional care. The four-month ramp isn’t a joke about CS being useless; it’s a measure of how much leverage the reasoning layer now has over the syntax layer. The labs figured this out first and are now doing talent arbitrage on it.

Where the premium moved

Main learnings

Reasoning beats syntax. The durable skill is judging the output, not producing the tokens.
Taste is the new technical skill. Knowing what good looks like is now literally the job description.
Fellowships are talent arbitrage. The labs are buying undervalued reasoning wherever it’s priced cheapest.

Practical takeaway — hire for reasoning quality, train for tools. The tools change quarterly; the reasoning doesn’t.

Abhijit’s post

The Builder

The $500 Reel Stack, Open-Sourced

Via @damkina7 — motion design on Claude Opus 5.5

Damkin open-sourced the whole motion-design stack for Claude Opus 5.5 — repos, the seek(t) architecture, and the write-up “Autonomous Motion Studio on Claude Opus 5.5: From seek(t) Architecture to Your First $500 Reel.” It’s the month’s repeating pattern: a frontier capability drops, someone turns it into an open stack, and the finish line is stated in revenue — your first $500 reel. The demo isn’t the render. The demo is the invoice. Damkin’s thread

The Briefs

The cold-email game, cracked

Sarvagya Kulshreshtha, quoting Sumit Kesarwani’s claim of cracking cold email — “works 8 out of 10 times” — with the driest possible endorsement: “same. works all the time.” The outbound channel isn’t dead; the generic version is. Sarvagya’s post

Cuck coding

Zam’s rebrand of the year: “call it cuck coding instead of vibe coding because you just sit in a chair and watch.” The most honest job description in software right now. Zam’s post

Muse, meet my credit card

Greg Hunkins: “Muse after I give it access to my credit card and the open internet.” The agentic-commerce trust leap, told as a joke that isn’t quite a joke. Greg’s post

Hogwarts, anime-accurate

XCaliber’s Dumbledore v. Tom Riddle Part 5 — the book-accurate version as anime. The fan-remake pipeline keeps industrialising; Hollywood’s IP is now everybody’s raw material. XCaliber’s video

The week’s quiet through-line

The demo is the product. A flying-car stunt, a trailer, a 14x burn multiple, a cold-email screenshot, an anime remake — everything this week arrived as a demonstration. Agents demo verification (Modal, ed22); this week the humans demoed everything else. The expensive thing wasn’t the intelligence. It was the showing.

Turns Out the Future Does Have Flying Cars

Main learnings

Inside this edition

The week’s through-line

The ₹168-Crore Househelp

Main learnings

India’s VFX Coming-Out Party

Main learnings

A Bubble in a Bubble

Main learnings

Philosophy Grads Teach the Code to Think

Main learnings

The $500 Reel Stack, Open-Sourced

The cold-email game, cracked

Cuck coding

Muse, meet my credit card

Hogwarts, anime-accurate

The week’s quiet through-line