Ep. #2: Secrets of Social Media Success, Mental Health & Distribution Hacks - 3 OGs Reveal All

Nikhil Kamath sits with Tanmay Bhat and two men who build social platforms for a living - Umang Bedi (Dailyhunt and Josh; ex-Facebook India MD) and Aprameya Radhakrishna (Koo; TaxiForSure founder) - for a nearly three-hour tour of how social media really works: the dopamine economics, the growth hacks, the creator money, and the mental-health bill. FULL COVERAGE: notes span the entire episode. Sponsor: none stated. All user numbers, revenue figures and anecdotes are as stated by the speakers in April 2023 and reflect that moment; nothing here is investment, medical or psychological advice. The mechanics first: platforms monetize attention - Umang's arithmetic has one user's ten daily minutes worth about $10 in ads, so a billion users is a billion dollars - which is why the incentive model is simply more users spending more time. The dopamine loop runs both ways, and Tanmay traces the exact inflection points: the retweet button and likes on comments turned audiences into performers. Growth wisdom from three operators: topicality is the number-one engagement lever (AIB kept memes ready for both PV Sindhu outcomes); every popular take creates a market for the contrarian one; people share relatable memes to be different. The money talk is unusually specific: Indian CPMs under $2 versus $10-20 in the US, why YouTube's 10-minute mid-roll change overnight stretched every video, Koo's plan to share 20-30% of revenue with users, and pay-for-performance creator funds on Josh (Faisal Shaikh's millions). Umang tells the WhatsApp-Taj-hotels targeting story as fact and lets you infer the rest, then lays out Dailyhunt's counter-design: no forced sign-in, no personal data, a content graph instead of a social graph, and deliberately shown diverse perspectives against confirmation bias. TikTok's India blitz gets numbers: $60-70 million burned per month, Facebook's largest global advertiser for two to three years, 240 million Indian MAUs at the ban - while China clocks 2.5 hours a day on short video and Saudi Arabia shows 98% penetration. The back half turns personal: Umang's 'most shallow human' reckoning at 40 (140 kilos, a title for an identity, a $100 million term sheet that vanished, one month of cash when the pandemic hit), Nikhil's 'be here now' tattoo math, and a raw segment on kids - live crimes on platforms, India's intermediary-accountability regulation, why Instagram ranks worst for how it makes people feel, and why monitoring beats banning. It closes with investment lenses: verticalization of social platforms, connected TVs in India, and why everyone at the table is quietly bullish on YouTube - plus Nikhil's Bill Gates breakfast, where Gates bet on fission because storage and transmission remain unsolved.

Executive Summary (TL;DR)

Three operators who build and feed social platforms explain the machine: attention is the product, envy is the fuel, and topicality is the growth hack. The numbers are concrete - $10 per user per ten minutes, sub-$2 Indian CPMs, TikTok's $70-million-a-month India burn, 240 million MAUs at the ban. The counters: Dailyhunt's no-sign-in, content-graph, anti-echo-chamber design; Koo's plan to pay users 20-30% of revenue. And the bill: mental health, kids' safety, and why the men running these platforms monitor rather than ban.

Key Takeaways

Attention has a price per minute

Umang's model: ten daily minutes of one user, one ad per foot of scroll, roughly ten dollars - scale to a billion users and the incentive model writes itself: more users, more time, more ad revenue. The dopamine hit that keeps you returning is the same hit that pays the platform, which is why every feature optimizes engagement over wellbeing.

Topicality beats everything

Tanmay's creator playbook from 2009 Twitter to AIB: whatever the country is talking about today - a Test match, a wedding, a final - will outperform anything evergreen. AIB kept memes ready for both PV Sindhu medal outcomes to cut posting latency to zero. Every popular sentiment then creates a market for the contrarian take, because people share relatable memes precisely to feel different.

The algorithm looks back at you

Umang on TikTok: when you look into it, it looks straight back - billions of untagged clips machine-labeled by AI, married to your implied preferences minute by minute, so no two feeds are alike. The numbers behind the machine: $60-70 million burned monthly in India, 240 million MAUs at the ban, and a content graph so effective that Douyin now drives 7% of all Chinese commerce.

A content graph instead of a social graph

Dailyhunt and Josh run on deliberate counter-design: never force sign-in, never ask for personal data, personalize on content adjacency rather than who your friends are, and - after the demonetization thought experiment - deliberately show diverse perspectives to fight confirmation bias. The admission that makes it credible: social platforms are inherently incentivized the other way.

Creators: India pays dimes, distribution is the moat

CPMs are sub-$2 in India versus $10-20 in the US; Instagram pays creators zero directly, so brand deals carry the economy. YouTube's mid-roll rules reshaped video lengths overnight, and its search-driven discovery fails local-language creators. Aprameya's response on Koo: share 20-30% of revenue with users themselves - his bet is India goes nuts the day a platform shows you what it earned off you.

The bill arrives as mental health

The panel is blunt: Instagram ranks worst for how it makes people feel (Umang's driver found it 'not aspirational but unattainable'), Twitter's mentions tab engineers negativity, and envy fuels both the hate comment and the clout farm. On kids, the consensus is monitoring over bans - linked accounts, watch history, and treating online perils as a new chapter in the good-touch-bad-touch talk.

Vertical, connected, and YouTube-bullish

Where the next decade goes: Tanmay bets on verticalized, purpose-specific platforms (the days of one-platform-serves-all are gone); Umang picks connected TVs as India's sleeper opportunity (only 200 million TVs for 1.4 billion people, and a TV-plus-fridge is the first aspirational purchase); everyone agrees print dies in ten years and that YouTube - with Learn and Podcasts - is quietly the strongest asset in the stack.

The Conversation Breakdown

Untimed topic

⏱ Untimed topic

The dopamine machine: likes as acknowledgement, the retweet-button inflection, the $10-per-user math, and a memory hack involving a thousand learned names

Untimed topic

⏱ Untimed topic

From Orkut walls to Facebook's handshake: how real identity won, the seven-friend magic number, and the aliases we all had

Untimed topic

⏱ Untimed topic

Umang's war stories: the WhatsApp-to-Facebook ad inference, Cambridge Analytica, the driver who found Instagram unreal, and Dailyhunt's four counter-principles

Untimed topic

⏱ Untimed topic

Creator economics laid bare: CPM gaps, mid-roll incentives, MrBeast's millions, Koo's pay-the-user plan, and why YouTube flows to every other platform

Untimed topic

⏱ Untimed topic

TikTok anatomy: the burn rate, the ban, China's 2.5 hours a day, Saudi's 98% penetration, and short video as micro-Bollywood

Core Frameworks & Ideas

The engagement ladder

Rank any content idea by rungs: topicality (today's event) first, then contrarian takes on the popular sentiment, then high positive sentiment (babies, dogs), then nostalgia. The paradox powering all of it: people share relatable things to signal they are different.

Content graph versus social graph

Two personalization philosophies: social graphs (who you follow) create echo chambers and confirmation bias; content graphs (what you actually watch, machine-tagged) can diversify on purpose - show fitness lovers dance, show both sides of an issue. The design choice determines whether the platform informs or polarizes.

The multi-dimensional network problem

A marketplace startup has two sides to subsidize; a social network has four - users, advertisers, creators, government - and all must be happy simultaneously, so growth cannot be bought with simple math. Network effects arrive serendipitously (a Twitter crisis, a TikTok ban); after takeoff, loyalty programs and creator monetization add the math.

Actionable Next Steps

✓Creators: pre-make your content for both outcomes of the next big event your audience cares about - posting latency is the real competition.
✓Audit your own feed this week: count how much is a social graph echo versus genuine discovery, and deliberately follow one strong source you disagree with.
✓If you have kids on devices, set up linked accounts and review watch history together - the panel's consensus is visibility and conversation over bans.
✓Building anything consumer? Write down which of the four stakeholders (users, advertisers, creators, regulators) your incentive model currently starves.

Notable Quotes

“when you're looking into Tik Tok Tik tok's looking straight back into you”
— Umang Bedi, on algorithmic personalization

“people share relatable memes to be different”
— Tanmay Bhat

“the world doesn't know how to treat nice people”
— Nikhil Kamath

The One Thing to Remember

Envy is the fuel of social media - the same loop that makes a follower feel inadequate makes a hater comment and a creator post; every platform decision, from the retweet button to the For You feed, is an attempt to refine that fuel.