A Parallel Elicitation-Free Protocol for Allocating Indivisible Goods

TL;DR

This paper introduces a parallel, elicitation-free protocol for allocating indivisible goods among agents, offering a promising alternative to sequential methods and analyzing strategic considerations.

Contribution

It proposes a novel parallel allocation protocol that is elicitation-free and compares its performance to traditional sequential methods.

Findings

Parallel protocol performs competitively with sequential methods.

The protocol is elicitation-free, reducing communication complexity.

Strategic behavior considerations are addressed.

Abstract

We study the problem of allocating a set of indivisible goods to multiple agents. Recent work [Bouveret and Lang, 2011] focused on allocating goods in a sequential way, and studied what is the "best" sequence of agents to pick objects based on utilitarian or egalitarian criterion. In this paper, we propose a parallel elicitation-free protocol for allocating indivisible goods. In every round of the allocation process, some agents will be selected (according to some policy) to report their preferred objects among those that remain, and every reported object will be allocated randomly to an agent reporting it. Empirical comparison between the parallel protocol (applying a simple selection policy) and the sequential protocol (applying the optimal sequence) reveals that our proposed protocol is promising. We also address strategical issues.