Abstract. In this paper, we address the problem of designing a robust reputation mechanism for peer-to-peer services. The mechanism we propose achieves high robustness against malicious peers (from individual or collusive ones) and provides incentive for participation. We show that the quality of the reputation value of trustworthy and participating peers is always better than the one of cheating and non participating ones. Finally we formally prove that, even when a high fraction of peers of the system exhibits a collusive behavior, a correct peer can still compute an accurate reputation mechanism towards a server, at the expense of a reasonable convergence time.

Recently, online reputation mechanisms have been proposed that reward agents for honest feedback about products and services with fixed quality. Many real-world settings, however, are inherently dynamic. As an example, consider a web service that wishes to publish the expected download speed of a file mirrored on different server sites. In contrast to the models of Miller, Resnick and Zeckhauser and of Jurca and Faltings, the quality of the service (e. g., a server’s available bandwidth) changes over time and future agents are solely interested in the present quality levels. We show that hidden Markov models (HMM) provide natural generalizations of these static models and design a payment scheme that elicits honest reports from the agents after they have experienced the quality of the service. 1

The most prominent way to establish trust between buyers and sellers on online auction sites are reputation mechanisms. Two drawbacks of this approach are the reliance on the seller being long-lived and the susceptibility to whitewashing. In this paper, we introduce so-called escrow mechanisms that avoid these problems by installing a trusted intermediary which forwards the payment to the seller only if the buyer acknowledges that the good arrived in the promised condition. We address the incentive issues that arise and design an escrow mechanism that is incentive compatible, efficient, interim individually rational and ex ante budget-balanced. In contrast to previous work on trust and reputation, our approach does not rely on knowing the sellers ’ cost functions or the distribution of buyer valuations.

Reputation-based trust models using statistical learning have been intensively studied for distributed systems where peers behave maliciously. However practical applications of such models in environments with both malicious and rational behaviors are still very little understood. This paper studies the relation between accuracy of a computational trust model and its ability to effectively enforce cooperation among rational agents. We provide theoretical results showing under which conditions cooperation emerges when using a trust learning algorithms with given accuracy and how cooperation can be still sustained while reducing cost and accuracy of those algorithms. We then verify and extend these theoretical results to a variety of settings involving honest, malicious and strategic players through extensive simulation. These results will enable a much more targeted, cost-effective and realistic design for decentralized trust management systems, such as needed for peer-to-peer systems and electronic commerce. 1.

This paper is available online at www.jtaer.com DOI: 10.4067/S0718-18762010000200005

In this paper we argue for the benefits of enabling the self-organization of virtual on-line communities, which today are mainly formed and operated by centrally managed web servers. However, self-organization requires community members to contribute themselves different types of resources (e.g. bandwidth, storage, etc.), as in the case of peer-to-peer (P2P) systems. Unfortunately, this cooperation cannot be taken for granted. To address this obstacle, we introduce the notion of a cross-layer incentive mechanism. The main idea is to encourage the contribution of low-level resources using social incentives generated at a higher (social) layer. We believe that this type of incentive mechanisms will play a central role toward the realization of self-organized virtual communities and will enable users to take advantage of the attractiveness and value of Web-based communities on the one hand and the externalities and flexibility of P2P networks on the other. We make a first step toward this direction: a) we categorize the different types of social incentives applicable in this context and b) we provide insights for the design of the appropriate social software required to map the behavior of participants at the resource sharing layer with suitable rewards at the social layer. 1.

This work introduces a probabilistic model of reputation allowing to compute reputation scores as close as possible to their intrinsic value, according to the model. It is based on the following, natural, consumer-provider interaction model. Consumers are assumed to order items from providers, who each has some intrinsic, latent, “quality of service” score. In the basic model, the providers supply the items with a quality following a normal law, centered on their intrinsic “quality of service”. The consumers, after the reception and the inspection of the item, rate it according to a linear function of its quality – a standard regression model. This regression model accounts for the bias of the consumer in providing ratings as well as his reactivity towards changes in item quality. Moreover, the constancy of the provider in supplying an equal quality level when delivering the items is estimated by the standard deviation of his normal law of item quality generation. Symmetrically, the consistency of the consumer in providing similar ratings for a given quality is quantified by the standard deviation of his normal law of ratings generation. Two extensions of this basic model are considered as well: a model accounting for truncation of the ratings and a

A well known problem in peer-to-peer overlays is that no single entity has control over the software, hardware and configuration of peers. Thus, each peer can selfishly adapt its behaviour to maximise its benefit from the overlay. This thesis is concerned with the modelling and design of incentive mechanisms for QoS-overlays: resource allocation protocols that provide strategic peers with participation incentives, while at the same time optimising the performance of the peer-to-peer distribution overlay. The contributions of this thesis are as follows. First, we present PledgeRoute, a novel contribution accounting system that can be used, along with a set of reciprocity policies, as an incentive mechanism to encourage peers to contribute resources even when users are not actively consuming overlay services. This mechanism uses a decentralised credit network, is resilient to sybil attacks, and allows peers to achieve time and space deferred contribution reciprocity. Then, we present a novel, QoS-aware resource allocation model based on Vickrey auctions that uses PledgeRoute as a substrate. It acts as an incentive mechanism by providing efficient overlay construction, while at the same time allocating increasing service quality to those peers that contribute more to the network. The model is then applied to lagsensitive chunk swarming, and some of its properties are explored for different peer delay distributions. When considering QoS overlays deployed over the best-effort Internet, the quality received by a client cannot be adjudicated completely to either its serving peer or the intervening network between them. By drawing parallels between this situation and well-known hidden action situations in microeconomics, we propose a novel scheme to ensure adherence to advertised QoS levels. We then apply it to delay-sensitive chunk distribution overlays and present the optimal contract payments required, along with a method for QoS contract enforcement through reciprocative strategies. We also present a probabilistic model for application-layer delay as a function of the prevailing network conditions. Finally, we address the incentives of managed overlays, and the prediction of their behaviour. We propose two novel models of multihoming managed overlay incentives in which overlays can freely allocate their traffic flows between different ISPs. One is obtained by optimising an overlay utility function with desired properties, while the other is designed for data-driven least-squares fitting of the cross elasticity of demand. This last model is then used to solve for ISP profit maximisation.

Almost every e-commerce site employs a so-called reputation mechanism that collects and publishes ratings from its users which then allow other market participants to make betterinformed choices. It is instructive to distinguish between two

Abstract. Though trust and reputation systems have been extensively studied, general architectural commonalities between the two have received little attention. In this paper, we present a life cycle model of reputation and trust systems, along with accompanying measures of how much effect signaling and sanctioning have on a given system. We map reputation attacks within our framework and apply our framework to an online auction model. 1