The Price of Ignorance: Information-Free Quotation Mechanisms for Data Retention in Machine Unlearning

Researchers propose an information-free quotation mechanism that requires no private user information for machine unlearning scenarios under the GDPR right to erasure. The server broadcasts increasing prices, and users choose their own data supply; the paper defines the welfare gap between this mechanism and optimal personalized pricing as the Price of Ignorance, and proves the efficiency ordering of three cases. In Monte Carlo simulations with 7 mechanisms and 5000 runs, the mechanism achieves ≥99% of the welfare of information-intensive benchmarks, with noise robustness and comparable fairness.