Parallel Contests for Crowdsourcing Reviews: Existence and Quality of Equilibria

Abstract

Motivated by the intricacies of allocating treasury funds in blockchain settings, we study the problem of crowdsourcing reviews for many different proposals, in parallel. During the reviewing phase, every reviewer can select the proposals to write reviews for, as well as the quality of each review. The quality levels follow certain very coarse community guidelines and can have values such as 'excellent' or 'good'. Based on these scores and the distribution of reviews, every reviewer will receive some reward for their efforts. In this paper, we design a reward scheme and show that it always has pure Nash equilibria, for any set of proposals and reviewers. In addition, we show that these equilibria guarantee constant factor approximations for two natural metrics: the total quality of all reviews, as well as the fraction of proposals that received at least one review, compared to the optimal outcome. * This work was supported by the ERC Advanced Grant 788893 AMDROMA "Algorithmic and Mechanism Design Research in Online Markets", and the MIUR PRIN project ALGADIMAR "Algorithms, Games, and Digital Markets".the necessary new features, with the final decision coming directly from aggregating user preferences. Some blockchains such as Cardano and Dash take this approach, having a publicly controlled treasury which allocates funds to proposals generated by the community. This process requires the careful design of many mechanisms, drawing from areas such as Crowdsourcing, Participatory Budgeting [4] and Distortion [1]. In this work, we focus on one important part of this process: how to provide the right incentives for the community to produce high quality reviews across many proposals.We provide a high level description of the relevant modules of Project Catalyst, the mechanism used by Cardano's treasury, as motivation for our modelling assumption and results. A new funding round takes place every 12 weeks. The latest, Fund6 allocated about $4, 000, 000 worth of ADA (Cardano's native currency), with $165, 000 given as rewards to reviewers. After the members of the community submit their proposals, there are 2 kinds of entities involved in the reviewing process: veteran community advisors (vCA's), community advisors (CA's). The CA's start by writing reviews for proposals. The reviews contain a written evaluation of the proposal, as well as a numerical score. The reviews (following some filtering for profanity, plagiarism, etc) are published immediately under the CA's pseudonym, while the true identity of the person is usually secret. Then, the vCA's write meta-reviews (i.e., they review the reviews submitted by the CA's). This evaluation is only on the quality of the review itself, not its score or subjective opinion. The evaluation follows specific, well-known guidelines and is very coarse: reviews can either be 'excellent', 'good' or 'filtered out' (which are discarded). The CA's are then rewarded based on the quality of their reviews, as well as choice of proposals to review, relative to actions of the other CA's.The re...

Simultaneous Contests with Equal Sharing Allocation of Prizes: Computational Complexity and Price of Anarchy

“…Two very recent related papers are by Birmpas et al [9] and Elkind et al [17]. Both these papers do not have activities, i.e., the players produce output directly for the contests, which makes their models a bit different (simpler) than ours, but they add complexity along other dimensions.…”

“…Therefore, their results are not directly comparable to ours. Birmpas et al [9] have both budgets and costs in the same model, and they give a constant factor PoA bound by augmenting players' budgets when computing the equilibrium welfare (but not when computing the optimal welfare). Elkind et al [17] consider a model with only one contest and in the case of incomplete information.…”