We study the trade-off between the Price of Anarchy (PoA) and the Price of Stability (PoS) in mechanism design, in the prototypical problem of unrelated machine scheduling. We give bounds on the space of feasible mechanisms with respect to the above metrics, and observe that two fundamental mechanisms, namely the First-Price (FP) and the Second-Price (SP), lie on the two opposite extrema of this boundary. Furthermore, for the natural class of anonymous task-independent mechanisms, we completely characterize the PoA/PoS Pareto frontier; we design a class of optimal mechanisms SP α that lie exactly on this frontier. In particular, these mechanisms range smoothly, with respect to parameter α ≥ 1 across the frontier, between the First-Price (SP 1 ) and Second-Price (SP ∞ ) mechanisms.En route to these results, we also provide a definitive answer to an important question related to the scheduling problem, namely whether non-truthful mechanisms can provide better makespan guarantees in the equilibrium, compared to truthful ones. We answer this question in the negative, by proving that the Price of Anarchy of all scheduling mechanisms is at least n, where n is the number of machines.property, an alternative approach to the framework of Nisan and Ronen [36] is to design mechanisms that perform well in the equilibrium, i.e., they provide good PoA or PoS guarantees. This approach has been adopted, among others, by central papers in the field (e.g., see [40,43] and references therein) and is by now as much a part of algorithmic mechanism design as the original framework of [36]. An interesting question that has arisen in many settings is whether non-truthful mechanisms (evaluated at the worst-case equilibrium, in terms of their PoA) can actually outperform truthful ones (evaluated at the truth-telling, dominant strategy equilibrium), for a given objective [11,20,24].While the literature that studies the concepts of PoA and PoS is long and extensive, there seems to be a lack of a systematic approach investigating the trade-off between the two notions simultaneously. More concretely, given a problem in algorithmic mechanism design, it seems quite natural to explore not only the best mechanisms in terms of the two notions independently, but also the mechanisms that achieve the best trade-off between the two. In a sense, this approach concerns a "tighter" optimality notion, as among a set of mechanisms with an "acceptable" Price of Anarchy guarantee, we would like to identify the ones that provide the best possible Price of Stability. Our main contribution in the current paper is the proposal of such a research agenda and its application on the canonical problem in the field, introduced in the seminal work of Nisan and Ronen [36], that of scheduling on unrelated machines.
Our contributionsPoA/PoS trade-off: We propose the research agenda of studying systematically the tradeoff between the Price of Anarchy and the Price of Stability in algorithmic mechanism design. Specifically, given a problem at hand and an objective function,...