Search
Search
Close this search box.

hu / en

Urban highways are barriers to social ties by Sándor Juhász Read more

Read more

Technology startups in Central and Eastern Europe: are they CEE-specific? Podcast by Andrea Szalavetz Read more

Read more

Socioeconomic inequalities and diabetes complications: an analysis of administrative data from Hungary - by Péter Elek, Balázs Mayer & Orsolya Varga Read more

Read more

Maximum-utility Popular Matchings with Bounded Instability - new research article by Ildikó Schlotter and Ágnes Cseh Read more

Read more

KTI Seminar: Christopher Stapenhurst – Randomized Deferred Acceptance and Possibility Based Strategy Proofness

The presentation will take place in a hybrid format via zoom interface or in person in the room K.0.11-12 on 29.05.2025, from 13.00.

Speaker: Christopher Stapenhurst (BME)

Title: Randomized Deferred Acceptance and Possibility Based Strategy Proofness
(joint work with József Pínter and Regina Stangl)

Abstract:

The deferred acceptance (DA) algorithm is known to be strategy-proof only for the proposing side in two-sided matching, with no deterministic and stable mechanism achieving strategy-proofness for both sides. We propose a randomized variant of DA—where man-proposing and woman-proposing DA are selected with equal probability—and show that it achieves strategy-proofness when agents evaluate lotteries over matchings optimistically (i.e., focusing on their best possible realization). To generalize this insight, we introduce possibility-based strategy-proofness (PBSP), a new incentive concept requiring that no agent can misreport preferences to make a strictly better matching possible in the lottery. We prove that random DA satisfies PBSP, offering a viable path to incentive-compatible stable matching without restricting to deterministic mechanisms. Our results demonstrate how stochasticity, combined with ordinal evaluations, can circumvent classic impossibility theorems.