Korea's secondhand apps had to manufacture the trust that made them work
Hyperlocal resale platforms grew by engineering assurance rather than matching supply and demand, and the design holds up well for a used rice cooker and much less well for a used phone.
Korea’s used-goods market was, for most of the internet era, a large online flea market run inside a web forum. Joonggonara, founded as a portal café in 2003, accumulated tens of millions of members and functioned as the default venue for selling a used monitor or a spare concert ticket. Industry estimates place the total secondhand trading market in the tens of trillions of won by the early 2020s, up several-fold over a decade — a figure assembled from platform disclosures and survey work rather than official statistics, and best read as a direction rather than a measurement.
The forum model worked and was miserable. Listings were posts, search was poor, and the central problem of any peer-to-peer market went entirely unaddressed: a buyer sending money to a stranger’s bank account had no recourse if nothing arrived. What changed the category was not better matching but a structural constraint that made deception expensive.
The hyperlocal apps, of which Danggeun Market — marketed abroad as Karrot — is the defining example, restrict what a user can see to listings within a few kilometres of a verified location. The constraint looks like a limitation and functions as a trust mechanism. A radius small enough to walk or drive across means transactions default to meeting in person and handing over cash, which removes the shipping-and-payment gap where most fraud lives. It also means the counterparty is a neighbour who may reappear at the same convenience store next week. Anonymity is what makes online markets cheap to defraud; proximity partially withdraws it.
On top of the geography the platforms layered explicit reputation. Danggeun’s “manner temperature” starts every user at 36.5 degrees and moves with transaction reviews, a legible single number in place of star ratings that most users ignore. Reputation scores are notoriously easy to inflate in anonymous markets, where accounts are free and disposable; tied to a verified neighbourhood and a history of physical meetings, the same score becomes moderately costly to fake. None of this is cryptographic assurance. It is friction, priced so that ordinary trades pass through and industrialised fraud does not.
The design has a clear boundary, and it is the price of the item. Nobody drives an hour to inspect a fifteen-thousand-won kettle, so cheap goods stay local and safe. Expensive goods — phones, cameras, luxury handbags, sneakers, appliances, event tickets — draw buyers from outside the radius, and those trades revert to remote payment and shipping, where the protective structure disappears. Korean police cybercrime statistics have recorded tens of thousands of reports of direct-transaction fraud annually for years, and the persistence of community-maintained blacklist services, where users look up a seller’s account number before paying, is evidence that platform-level assurance did not reach the transactions that hurt.
The platforms’ answer has been to add back the intermediation they originally avoided. Escrow-style payment services hold funds until the buyer confirms receipt; authentication marketplaces for sneakers and luxury goods take physical custody, inspect the item and ship it on, which is a fundamentally different and much more expensive business. Both work. Both are optional in the peer-to-peer case, and optional protection is declined precisely by the users in a hurry to close a deal, which is the population fraud selects for.
There is an economic footnote worth keeping. Because the hyperlocal model was built on free local trades, its operator took no cut of transactions and lost money for years, reporting its first full-year operating profit only in 2023, on advertising to local businesses rather than on the trading itself. The trust apparatus — verification, reputation, moderation, dispute handling — is pure cost. That is the awkward part of the lesson: what makes these markets function is the part nobody is paying for directly.