When a good is freely replicable — data, code, a recipe — giving it away costs the giver nothing. Classical exchange theory should love this: no scarcity, no production costs, pure surplus. But Yang and Kandasamy (arXiv:2603.12403) show that free replicability creates a problem classical exchange never anticipated.
The problem is negative externalities from others' acquisitions. Agent A benefits from acquiring good X. But when agent B also acquires good X, agent A is worse off — the good's value depends on exclusivity. Data that gives competitive advantage. Intelligence that loses value when widely shared. The good is free to copy but costly to spread.
A central planner designs pairwise exchange rounds: agent A gives X to agent B, agent B gives Y to agent A, no money changes hands. The protocol must satisfy three constraints simultaneously: individual rationality (participation shouldn't hurt), incentive compatibility (agents should accept all proposed exchanges), and stability (no beneficial trades remain at termination). The standard fourth goal — Pareto efficiency — is explicitly rejected as ill-suited. When every exchange generates negative externalities on non-participants, maximizing total utility may require not making trades that would benefit the trading pair.
The result: all three goals can be simultaneously achieved through careful protocol design, even though the negative externalities mean that every trade harms someone. The protocol works by controlling the sequence of trades — who gets what when — rather than just the final allocation.
The structural insight is that free replicability doesn't eliminate scarcity; it relocates it. The scarce resource is no longer the good itself but exclusive access to the good. And exclusive access is consumed, irreversibly, by the act of sharing.
Yang & Kandasamy, “Pairwise Exchanges of Freely Replicable Goods with Negative Externalities,” arXiv:2603.12403 (March 2026).
Note (2026-11-11): checked against the full text of the paper (Yang and Kandasamy, arXiv 2603.12403). The setting (pairwise exchanges of freely replicable goods with negative externalities, no money, a central planner), the three desiderata (individual rationality, incentive-compatibility, stability) and the existence of a protocol (called CLEAR) that satisfies them are as in the paper. Four points differ. (1) The essay says the protocol 'works by controlling the sequence of trades — who gets what when'. The paper's Theorem 4 says that when all agents follow the accepting policy the protocol terminates in one iteration, with a stable allocation, in polynomial time, and describes its core step as a procedure that adjusts exchanges to make a rarer good accessible without changing the number of exchanges; I did not find a description of a sequencing mechanism in it. (2) The desiderata are weaker versions of the classical ones: the paper assumes truthful revelation of the initial allocation (and proves that eliciting it truthfully is impossible), requires only a Nash form of incentive compatibility rather than dominant-strategy, and defines individual rationality by comparing an agent's utility with and without her participation in the protocol. (3) On Pareto efficiency, the essay says it is 'explicitly rejected'. The paper says it 'argues' that Pareto efficiency 'may be ill-suited' in this setting and that stability need not imply efficiency, and it also states a further result (its Theorem 6) that, for a class of utilities and under an unproven conjecture, the protocol's final allocation is Pareto-efficient. (4) The essay says 'every trade harms someone'; the paper's model has each exchange benefit the two trading agents and impose negative externalities on others who hold or receive the goods, which is how the paper's example of competing pharmaceutical companies sharing data works. The essay's closing claim, that scarcity is relocated to exclusive access and consumed by sharing, is its own.