Incentive compatible mechanism for trust revelation
Incentive compatible mechanism for trust revelation
复制标题
信任揭示的激励兼容机制
DOI：
--
复制
发表时间：
2002
期刊：
影响因子：
--
通讯作者：
T. Sandholm
中科院分区：
文献类型：
--
作者：
S. Braynov;T. Sandholm
Most of the work on trust is build around the assumption that agents entertain beliefs or estimates of other agents’ trustworthiness. Such beliefs help agents make decisions and reason about other agents’ trustworthiness. Obtaining and maintaining trust beliefs and estimates, however, is a serious practical problem for which no satisfactory solution has been found. Trust assessment and evaluation usually runs into the following problems. First, trust learning requires long-term interaction and is usually costly for the learning agent who has to accept the risk of being abused for learning purposes. Learning costs may include information search costs, costs for obtaining additional guarantees from trusted third parties, etc. If the costs are prohibitively high, then an interaction may fail, regardless of the trustworthiness of the other party. Second, trust typically is learned gradually, but can be destroyed in an instant by misfortune or a mistake. Once trust is lost, it may be costly or it may take a long time to rebuild it. This reflects certain fundamental mechanisms of human psychology known as the asymmetry principle [6]. Third, the process of trust learning seldom produces complete and accurate estimates. Inaccurate beliefs could lead to interaction failures and inefficiencies. For example, an agent with an inaccurate estimate of his partner’s trustworthiness might decide not to participate in an interaction, even when the other party is completely trustworthy. In our previous research [1] we analyzed the impact of trust on market efficiency, and we showed that the accuracy of trust beliefs and estimates is a crucial factor for market efficiency. We proved that inaccurate trust estimates reduce
