I am back to building. The idea is a marketplace, born out of a frustrating personal experience with incumbents who are large, visible, and slowly dying. Here is what re-reading Andrew Chen’s The Cold Start Problem taught me the second time around.
Re-reading a book a second time is underrated. When I first read Andrew Chen’s The Cold Start Problem years ago, I read it as a product builder who admired marketplaces from the outside. Reading it now, as someone actually trying to build one, the same sentences carry different weight. The frameworks have not changed. What has changed is that I now know which parts of them are load-bearing and which parts I used to nod at without really understanding.
The idea I am building is informed by a gap I experience personally. The incumbents in the category are large, reasonably well-funded, and deeply unprofitable. They have been in market for years. On paper, they look like winners. In the transaction itself, they fail their users consistently enough that I and almost everyone I know has at least one story. This is the gap. The question is how to exploit it, and whether what I am seeing is a real opening or a graveyard that has already swallowed better-funded attempts.
Re-reading the book, three ideas from outside the tech canon kept returning to me. Each describes, in a different discipline, the same phenomenon that governs marketplaces.
Three Frameworks, One Phenomenon
In the 1930s, a University of Chicago zoologist named Warder Clyde Allee made an observation so counterintuitive, it took decades to be taken seriously. Through careful experiments with goldfish, he demonstrated that social animals survive more readily in groups than in isolation, not because of safety in numbers, but because cooperation itself generates fitness. More critically, he identified a minimum viable group size below which a population does not merely grow slowly, but begins to die. The scarcity itself becomes lethal. He called this the Allee threshold: a point of no return, below which the system spirals and above which it compounds.
Decades later, Robert Metcalfe, co-inventor of Ethernet, articulated the complementary idea for networks: their value grows proportional to the square of connected users. Each new participant adds value not just for themselves but for every existing member simultaneously. WhatsApp with two billion users is not a thousand times more valuable than WhatsApp with two million, it is exponentially more so. Size becomes a structural advantage.
Then there is escape velocity, the speed required to break free of a gravitational body’s pull. Below it, gravity wins. Above it, momentum sustains itself. This is the underlying physics of Amazon’s flywheel and, honestly, the reason I named this blog what I did. For founders, it is the most honest metaphor we have.
Three frameworks, three disciplines, one phenomenon: systems on either side of a critical threshold where one side leads to collapse and the other to compounding. For marketplace builders, this is foundational. But it is incomplete. These frameworks explain how to cross the threshold. They do not explain what happens afterward. And “what happens afterward” is where most incumbents quietly begin to die while still appearing to be alive.
The incumbents I am competing with already crossed the threshold. They are large. They are still dying. Understanding why is the entire game.
The Cold Start Problem
Every two-sided marketplace faces the same founding problem: supply has no reason to join without demand, and demand has no reason to join without supply. You have two primary paths out of this. The first is to raise significant capital and incentivize both sides heavily until the flywheel spins on its own, worrying about unit economics only after escape velocity. The second is to find a complementary partner who can guarantee one side, freeing you to concentrate all your effort on the other.
When it comes to choosing the initial micro-market to test your hypothesis, you could either choose the largest with most liquidity but also the most fiercely competitive or one where the incumbents are the weakest. I learnt that, DoorDash famously chose suburbs rather than dense cities, not because suburbs were sexier, but because the incumbent was weakest there and the Allee threshold was lowest. The goal is not to plant a flag in the most visible market. The goal is to reach escape velocity before incumbents take notice. The strategic choices I made, didn’t allow me to choose a suburban micro-market but I did choose a wedge where incumbent isn’t yet focussed.
Bad Supply, Good Supply
Andrew Chen suggests, here is where the standard narrative breaks down. Metcalfe’s Law measures potential connections between nodes. It says nothing about the quality of interactions along those connections. This is its most serious shortcoming as applied to marketplaces. The fourth idea comes from the 16th century, Thomas Gresham’s law applied to marketplaces, which infers that bad supply drives out good supply. If a marketplace cannot distinguish quality, low-quality participants flood in because the barriers are low. Quality perception deteriorates. High-quality supply, which depends on quality-seeking demand, begins to exit. The network continues to grow in nominal size. Its unit economics quietly rot.
This is exactly what is happening to the incumbents I am circling. eBay has fought counterfeit goods for decades without winning. Fiverr’s bidding culture drives pricing toward zero, pushing premium freelancers out and attracting buyers who only want cheap. Zillow in the US and Magicbricks in India suffer from fake listings planted by agents to capture leads; supply that exists not to transact but to farm contacts. In each case, the platform looks large from the outside. From inside the transaction, it is broken. And the longer the platform tolerates it, the more the good supply exits and is replaced by supply that depends on the dysfunction.
Trust mechanisms are not product features. They are the product.
Airbnb understood this early. Guest identity verification and the Host Guarantee — which insured hosts against property damage, were not marketing gimmicks. They were core product infrastructure. The Host Guarantee cost real money to operate. It also made high-quality hosts willing to list their best properties rather than their spare rooms, which transformed the supply quality of the entire network, which made demand willing to pay more, which made high-quality hosts more profitable, which attracted more of them. Trust infrastructure created a compounding effect on quality that paralleled the compounding effect of scale. Most marketplaces pick one. The durable ones pick both.
The second hidden variable is matching efficiency. Two-sided density is necessary but not sufficient. A marketplace with significant supply and significant demand but weak matching is a marketplace of frustrated participants. Density without curation is noise, not liquidity. This is the second failure mode I see in the incumbent I am competing with: they have volume, but the act of finding what you actually want inside that volume is exhausting. Scale without matching is not an asset. It is evidence of neglect.
Crossing the Allee threshold and reaching escape velocity creates a problem the threshold frameworks do not anticipate: Power. A marketplace above escape velocity has significant leverage over the supply and demand side. How that power is exercised determines whether momentum sustains itself or creates a gap for a disruptor.
The Indian market is experiencing this right now. Swiggy and Zomato in food delivery, and Blinkit and Zepto in quick commerce, extract significant commissions from the restaurant and seller supply side. Uber and Ola did that too, until Namma Yatri and Rapido, charging drivers a small daily subscription fee instead of per-ride commission, forced Uber and Ola to fundamentally rethink their model. In doing so, Rapido exposed the fragility of extractive economics at scale. Swiggy/Zomato and Blinktin/Zepto are yet to experience the same. When the alternative exists and is credible, the platform that extracts discovers that its participants were never as captive as its valuation assumed. In each case, the platform confused its escape velocity for permanent immunity. The network is large, so the Allee effect feels irrelevant. But erode supply quality or supply willingness far enough, and the threshold becomes relevant again, on it’s way down. Compounding runs in reverse. This is what I am hoping would play in my favour, in the industry that I picked.
The Unified Framework
My current working mental model: every marketplace passes through three distinct phases, each with its own governing logic and its own way of dying.
Phase 1: Below the Allee Threshold
Strategic choices are about concentrating enough participants in a small enough space to make the network self-sustaining.
Failure mode is spreading too thin in such a large micro-market that you fail to achieve required density for self-sustenance.
Phase 2: Crossing Escape Velocity
What trust infrastructure do you need to build to prevent Gresham's law from kicking in and degrading your supply. The failure mode here is treating scale as the only relevant metric while the network rots from within.
Phase 3 · Above Escape Velocity
How do you exercise the power, network scale has given you in a way that keeps your highest-value participants committed, rather than motivated to build alternatives? Failure mode: the extractive platform that confuses current dominance for permanent safety, compresses supply economics until quality erodes, and discovers the Allee threshold again on the way down.
Allee, Metcalfe & Gresham weren’t thinking of two-sided marketplaces, but their principles that cooperation creates value, value compounds with scale, bad quality destroys good quality if left unchecked – are underlying principles of complex marketplaces.
The three questions I keep returning to, and the questions I will keep asking myself as I build:
- Where is our threshold, and how do we cross it?
- What trust infrastructure do we bake in from day one so that what compounds is actually valuable?
- And if we do reach escape velocity, how do we govern the power that scale affords us without turning it against the people who gave it to us in the first place?
I do not have complete answers yet. I have the questions. For now, that feels like enough to start.
The image was generated using Gemini AI, while the background research and light text editing was done using Claude AI. This writing is to help crystalize my thoughts, more than preaching to the choir.