General utilities and optimistic-agent decoy design

Extend the polling framework for welfare-maximizing intervention in multi-activity network games beyond linear-quadratic utility functions and design decoy signals that induce truthful reports from agents with more optimistic ambiguity attitudes.

Background

The paper studies a polling algorithm for learning welfare-maximizing non-discriminatory prices in multi-activity network games with heterogeneous private technologies. The theoretical characterization of optimal intervention and the convergence analysis rely on a linear-quadratic utility specification, while the payment-free incentive guarantee relies on agents exhibiting sufficient ambiguity aversion.

The paper also develops decoy-signal mechanisms that can induce approximately truthful reporting by ambiguity-averse agents and, under suitable conditions, by agents with intermediate theta-maxmin preferences. However, the authors do not establish an extension to general utility functions or a decoy construction that works for more optimistic agents. These extensions are therefore explicitly identified as unresolved.