New auction design lets self-funded marketplaces form without external budgets

Computer scientists from Google and academia just dropped a paper that could reshape how decentralized platforms and marketplaces are bootstrapped. Titled "Forging Self-Funded Marketplaces among Strategic Agents," the work (arXiv:2608.14548, accepted at ACM EC 2026) tackles a fundamental problem: how to get rational, self-interested agents to contribute effort to build a marketplace when there's no external budget to pay them.

The model is elegantly simple. Each agent i picks an effort level x_i, suffers a private cost c_i (unknown to the designer), and generates revenue r_i proportional to that effort. Critically, c_i can be larger or smaller than r_i—some agents are net contributors, others are net drains. The mechanism must choose an effort profile to maximize total value, while guaranteeing each agent is compensated at least their cost, and that total payments never exceed total generated revenue. This is budget-balance with a twist: the budget itself is endogenous, created by the agents' own efforts.

The authors show this problem is fundamentally harder than classic budget-feasible mechanism design. They prove that no truthful auction can achieve a bounded approximation of the first-best outcome, and even in restricted cases, the best you can do is logarithmic. To beat that, they propose a class of sequential auctions whose subgame perfect equilibria achieve a logarithmic approximation—matching the lower bound. Then they introduce the maximin share (MMS) benchmark, which captures "market thickness"—how much value can be extracted when agents coordinate—and deliver an auction that achieves a constant approximation of that benchmark.

The takeaway: if you want a marketplace to fund its own creation, you need to think in terms of sequential interaction and equilibrium refinement, not one-shot truthful auctions. This has immediate implications for DAOs, open-source funding, and any network where participants collectively create the value they're rewarded from.