ITR/AP:Computer Test Beds and Mechanism Design

机制允许信息分散的个人实现期望的群体结果。实验证实了理论预测，不同的参数值和不同的机制规则会对结果的效率和收敛速度产生重要影响。将开发一个基于计算机的行为模型，以测试广泛的机制和机构。该模型将基于进化算法。每个智能代理都有一组不同的策略可在其决策过程中使用。战略通过基于对每个战略的假设和实际绩效的评估的选择过程来竞争做出代理决策的权利。该模型将首先通过有重要实验数据的机制进行调整。例如，格罗夫斯-莱德亚德机制、公共产品自愿捐助机制、双向拍卖和在一组平行市场上的同时交易。然后，该模型将被用来预测那些尚未在人类受试者身上进行实验测试的机制下的行为。这个项目将开发一种新的经济学研究方法，有可能比实验经济学更好、更快、更便宜。

Mechanisms allow individuals with dispersed information to achieve desired group outcomes. Experiments confirm theoretical predictions that different parameter values and different mechanism rules can have important effects on the efficiency of outcomes and the speed of convergence. A computer-based behavioral model will be developed to test a wide range of mechanisms and institutions. The model is to be based on evolutionary algorithms. Each smart agent has a collection of different strategies to use in its decision-making processes. Strategies compete for the right to make agent decisions through a selection process based on an evaluation of each strategy's hypothetical and actual performance. The model will first be tuned with mechanisms for which there are significant experimental data. Examples include the Groves-Ledyard mechanism, the voluntary contribution mechanism for public goods, the double auction, and simultaneous trading in a set of parallel markets. Then the model will be used to make predictions about behavior under those mechanisms that have not already been tested in experiments with human subjects. This project will develop a new methodology for economics research that has potential to be better, faster, and cheaper than experimental economics.