|Abstract: ||「良善社會」，或是儒家所說之大同世界，乃是全人類的共同理想。對它的定義、憧憬，古今中外，已有如瀚海般之討論。然而，對於「良善社會」的「複雜性」，則少有研究。這使得我們常依賴「由上而下」的機制來建構「良善社會」，而無法透視其「由下而上」的「突發性」，這也使得我們無法對一些看似無關緊要，但卻有關鍵效果的隱微事件，做出適當的觀測和反應。 「計算社會科學」，或稱為「代理人基社會科學」或「社會模擬」，目前已成為美國國家科會基金會「師生創新科技經驗計畫」(ITEST) 中的一環，它不但能模擬出許多「隱微之處」，更能評估出他們的作用和影響。它配合真人實驗所發展的雙軌系統，已成為研究複雜科學的新典範。本計畫提出了影響「良善社會」複雜性的五大要素：社會偏好、社會信任、社會網路、社會智慧、以及社會規範，並藉由代理人基模型所建構的四個人工社會，配合實驗經濟學的方法，來了解這五大要素的整體運作如何促成「良善社會」的複雜性。 這四個人工社會中的前三個，是帶有賽局色彩的按艾法洛酒吧問題、施惠者—受惠者賽局、和網路信任賽局。這樣的選擇是由於它們簡單的結構，可以勾勒出一個具體而微的「良善社會」；然而，這樣的「良善社會」在現有文獻中，要不就尚未出現（像是艾法洛酒吧問題和網路信任賽局模型），要不就需要依賴極度嚴謹的數學操控始能達成（施惠者—受惠者賽局模型）。換句話說，「良善社會」的形成，看似簡單，實難以了解。因此，研究這五大要素的交互作用，正可讓我們了解「良善社會」複雜之所在。 第四個人工社會是預測市場。我們選擇預測市場的動機，來自於海耶克資訊加總機制（或稱海耶克假說），因為究竟資訊加總是集「眾人之痴」抑或是集「眾人之智」，本身即攸關社會災難的形成或避免，而此自然影響到「良善社會」之形成與發展。然而目前對預測市場的研究，多著重其在預測上的表現，少有鞭辟入裡的去研究從公民參與到集成智慧之間的複雜連結。本計畫將用代理人基模型及真人實驗，以社會網路為起點，逐漸加入社會規範、偏好與信任，來研究公民參與對集成智慧良窳之影響。|
Good society is a common dream for different people, regardless of their backgrounds and social status. Various descriptions of what a good society is have been proposed fromacademics to practitioners. A fundamental question which, however, has not been so much addressed is the complexity of a good society, regardless of the version of the complexity being applied. The general lack of the inquiries into the complexity issuemay cause us not being able to see the bottom-up emergent property of a good society and make us depend only on the top-down force to draw a blueprint which cannot foresee or react to many seemingly negligible fine details. Those initially small details are, nonetheless, powerful enough to deviate us, dramatically and permanently, from our targeted path. Now being part of the U.S. National Science Foundation’s Innovative Technology Experiences for Students and Teachers Program (ITEST), the recently rising computational social science, alternatively known as agent-based social science or social simulation, can not only generate and simulate many of these details, but also help us evaluate their impacts. This tool or platform can facilitate the blueprint design of a good society. In this integrated research project, five non-exhaustive essential elements which contribute to the complexity of a good society are identified as social preferences, social trust, social networks, social intelligence, and social norms, or simply, “Five Big Ss”. Through the study of four artificial societies, we hope to learn how these five elements can be synthesized into the agent-based modeling and simulation, coupled with human-subject experimental studies, of a good society. To have an initiate step toward the agent-based modeling and simulation of good societies, we start with three game-theoretic environments, namely, the El-Farol Bar problem, the donor-recipient game, and the network-based trust game. These three theoretical environments are chosen mainly for three considerations. Firstly, their high noticeability in the existing literature signifies their theoretical attractions, i.e., their structures are simple but have the power to construct the epitome of a good society. This advantage is related to the next consideration. Secondly, one can easily develop an intuitive but non-trivial notion of a good society in these three environments. In fact, even though the then defined good society is so non-trivial, in the existing literature, it either has never been substantiated (for the El Farol Bar case and for the network-based trust game) or it can be achieved only under very stringentmathematicalmanipulations (for the donor-recipient case). In other words, up to the present, few know the existence of the blueprints of good societies under these three theoretical environments. Thirdly, accordingly, the artificial societies built upon these three theoretical environments can be intriguingly enough to allow us to see the interplays of Five Big Ss in the complexity and in the formation of a good society. Our fourth artificial society is prediction market. There are two considerations for this choice. Firstly, the involvement of the prediction market is mainly motivated by Hayek’s information aggregation, or known as Hayek hypothesis, and its long-lasting influence to experimental economics. Secondly, we consider the prediction market itself as an important constituent of a good society if it can help predict the coming of a social disaster and prevent the society from catastrophic attacks. Unlike the previous three artificial societies, one can have an easy characterization of a good societywithin this environment, but the current literature on prediction market emphasize very much on its empirical performance instead of its theoretical underpinnings. What has been particularly ignored is the process from information distribution or dissemination to information aggregation under various social settings. Since literature is not clear in this part, it makes our research important, aiming to substantiate a good society in this environment.