Decentralised Norm Monitoring in Open Multi-Agent Systems: (Extended Abstract)

We propose an approach to norm monitoring in open multi-agents systems (MAS) in which monitoring is performed by the agents comprising the MAS. Using ideas from scrip systems, we show how to design mechanisms where agents are incentivised to monitor the actions of other agents for norm violations. Crtically, the cost of providing incentives is not borne by the MAS. Instead, monitoring incentives come from (scrip) fees for accessing the services provided by the MAS. In some cases perfect monitoring (and hence enforcement) can be achieved: no norms will be violated in equilibrium. In other cases, we show that, while it is impossible to achieve perfect enforcement, we can get arbitrarily close; we can make the probability of a norm violation in equilibrium arbitrarily small. 1. INCENTIVISING MONITORING We propose an approach to norm monitoring in open multi-agents systems (MAS) in which the monitoring of agent actions is performed by the agents comprising the MAS. We focus on norms which prohibit certain actions (or the state resulting from an action). The novelty of our approach is that the MAS does not bear the cost of monitoring; nor does it levy fines on violating agents to pay for monitoring as in, for example, [1]. (Levying fines is not possible in many open systems as the agents can always leave the system and rejoin later under a different identity.) Instead, monitoring incentives come from (scrip) fees [2] for accessing the services provided by the MAS. As a simple example, consider a MAS where agents want to post content on the web. There are norms regarding what may be posted; for example, copyrighted images should not be posted, and comments should not be abusive or defamatory. We assume that agents may occasionally submit posts that violate the norm. If such content appears on the web, the MAS loses significant utility (e.g., it can be fined or sued). It is therefore in the MAS’s interest that submitted posts are checked for compliance with the norm ∗Supported in part by NSF grants IIS-0534064, IIS-0812045, IIS0911036, and CCF-1214844, and by AFOSR grants FA9550-08-10438, FA9550-09-1-0266, and FA9550-12-1-0040, and ARO grant W911NF-09-1-0281. Appears in: Proceedings of the 15th International Conference on Autonomous Agents and Multiagent Systems (AAMAS 2016), J. Thangarajah, K. Tuyls, C. Jonker, S. Marsella (eds.), May 9–13, 2016, Singapore. Copyright c © 2016, International Foundation for Autonomous Agents and Multiagent Systems (www.ifaamas.org). All rights reserved. before they appear on the web. We assume that it is possible to check objectively if a particular item of content violates the norm. (For simplicity, we assume that if a post that violates the norm is checked, the violation will be detected. We can easily modify our approach to handle the case where there is some probability ρ of the violation being caught.) Checking whether a post is ‘bad’ (violates the norm) requires some work, and incurs a small utility cost. Although checking requires some resources, we assume that if a violation is found, evidence of the violation can be provided that can be checked in negligible time (so we do not need to deal with disputes about whether content violates the norm). If the content does violate the norm, the post is discarded and no violation occurs. We would like to distribute the monitoring of posts among the agents that use the system. Just as for the MAS, monitoring incurs a small negative utility for an agent. This means that agents must be appropriately incentivised to monitor. The scenario we describe applies to a wide range of normative multi-agent systems. The only requirement is that norms apply to single actions, actions can be checked in advance, and their execution prevented if they are found to violate the norm. Examples other than web postings include bids for jobs and bookings of facilities. We formalise the posting and monitoring of content for norm violations as a non-cooperative game. This scenario (and the resulting game) is similar to that considered by Friedman et al. [2], but differs in several key respects. As in [2], we adopt the idea of using tokens as payment for posting and as a reward for monitoring. 2. UNINTENTIONAL VIOLATION We first consider a scenario in which bad posts are unintentional, and happen with a constant probability b. In this scenario, agents are unaware that they are violating the norm when they post something inappropriate. The game in the inadvertent scenario is described by the following parameters: • a finite set of n agents 1, . . . , n; • the time between rounds is 1/n; • at each round t an agent is picked at random to submit a post (we implicitly assume that agents always have something that they want to post); • probability of a post being bad: b; • utility of posting (to the agent doing the posting): 1 • disutility of monitoring (to the agent doing the monitoring): −α (where 0 < α < 1); • discount rate: δ ∈ (0, 1).