Entropy and Marketplaces
Uber launched in multiple African cities between 2013 and 2016.1 Drivers moved their life savings into buying cars. Demand showed up. Ten years later, those same cars, the ones that were supposed to replace old, inefficient taxis, have themselves become old and inefficient. The disruption decayed back to the starting point.
I studied chemical engineering. The concept I keep returning to in business is entropy: real change generates disorder. You cannot rearrange a system without disturbing it. The useful distinction is between disorder that can be recovered and disorder that permanently destroys the resources the system needs to keep working.
Marketplaces generate entropy by design. They change how people move, pay, buy, borrow, and trust. They break old routes and build new ones. They ask suppliers to buy assets, customers to try unfamiliar behavior, regulators to tolerate new patterns, and investors to fund the transition before the system proves itself. The question is whether the disorder becomes infrastructure or waste.
The burned asset
Uber in Africa created irreversible entropy on the supply side. Drivers put in capital. Cars entered the system. Demand showed up. But there was no financing structure to renew the asset base, no data loop that made the vehicle scoreable, no bank standing behind the next cycle. The marketplace consumed driver savings and produced trips, but not enough durable capital formation.
That is the hard version of marketplace failure. The app can work and the economic structure can still decay. Riders get a better service for a while. Drivers earn for a while. Investors can point to volume. Then the vehicle stock ages, maintenance rises, driver economics weaken, and the system drifts back toward the condition it was meant to replace.
The same pattern appears in softer forms. A food-delivery market can train restaurants to depend on subsidized orders before delivery density exists. A lending marketplace can push credit into merchants before collections discipline is real. A logistics startup can sign customers before routes, warehouses, and addresses can carry the promised service. The disorder is visible as growth. The question is what survives after the subsidy or novelty leaves.
The political cost
Europe shows a different kind of entropy. There the asset base was not the primary failure. The disorder landed politically. Taxi unions, city governments, labor courts, safety regulators, data authorities. Each market fought the platform through a different institutional channel, and the accumulated resistance became a cost the business still pays.
Regulatory goodwill is a resource. So is public trust. Burn either one carelessly and the market may still function, but every later move becomes more expensive. A company that treats regulation as an obstacle rather than a system to be sequenced eventually spends its best years litigating what it could have designed around.
This is where emerging markets can be misunderstood. Weak formal institutions often hide stronger institutions elsewhere: unions, transport associations, municipal gatekeepers, market queens, agent networks, family capital, informal lenders. Disrupt the formal layer and you may face little resistance. Disrupt the actual layer and the city pushes back.
Thin disruption
James Scott called the good version a "thin simplification", an intervention that adds just enough structure to release value without destroying the local knowledge that makes the system work.2 That is the marketplace design problem in one sentence. Add enough legibility for matching, pricing, settlement, and financing. Avoid stripping out the operating knowledge that keeps the city moving.
A platform that instruments a transport company without replacing it is thin disruption. The company still manages drivers, maintenance, collections, and local relationships. The platform adds demand, routing, payments, and data. That data then makes the fleet financeable through The Financing Loop. The old system remains in place, but becomes more productive.
The opposite is thick disruption: replacing every local layer at once. Own the drivers directly. Own the vehicles. Own the warehouses. Own the customer relationship. Own the credit risk. That can work in some markets, but in most emerging cities it means absorbing every source of disorder onto one balance sheet. The company becomes the shock absorber for the whole city.
The sequence
The practical question is sequencing. Which side of the marketplace do you build first, and how do you connect each phase to the next so the resources invested in one cycle carry into the next?
In transport, this means supply first. Vehicles on roads, connected to demand, connected to data, connected to financing. The first cycle creates trips. The second creates an underwriting record. The third creates cheaper capital. Each cycle has to conserve something from the prior one. Otherwise the system only burns through more cars, more driver savings, and more promotional spend.
In commerce, the sequence changes. Payment trust, delivery trust, addressability, return behavior, and merchant discipline may all need to exist before the marketplace can carry physical goods. That is why Rails Before Marketplaces matters. Across marketplaces, the first question is which one the city's existing rails can carry. Inside a marketplace, the question is which side creates the least destructive disorder when it moves first.
The durable default
The entropy lens is useful because it separates volume from durability. A marketplace can grow by pushing disorder into suppliers, customers, regulators, or investors. That growth looks real until the resource being consumed runs out. Driver savings. Restaurant margins. Regulatory patience. Venture capital. Public trust.
The durable marketplace rearranges the city in a way that leaves behind a stronger system than it found. Vehicles renew instead of decay. Drivers become underwritable instead of exhausted. Merchants gain distribution without losing their local knowledge. Customers repeat because the service has become the easiest way to meet a daily need.
The businesses that last are the ones that fill a gap so fundamental that the disruption becomes the new default. Low entropy can still carry large ambition when the system keeps enough of its own resources to compound after the first shock.