作者: David Wolpert
DOI: 10.1007/978-1-4419-8909-3_2
关键词: Collective intelligence 、 Management science 、 BaseBean 、 Set (psychology) 、 Heuristics 、 Mathematics 、 Nash equilibrium 、 Construct (philosophy) 、 Function (engineering) 、 Structure (mathematical logic)
摘要: In this chapter an analysis of the behavior arbitrary (perhaps massive) collective computational processes in terms associated “world” utility function is presented We concentrate on situation where each process can be viewed as though it were striving to maximize its own private function. For such situations central design issue how initialize and update collective's structure, particular functions, so induce overall behave a way that has large values world utility. Traditional “team game” approaches problem simply set equal The “collective intelligence” (COIN) framework semiformal heuristics have recently been used construct functions many experiments resulted up orders magnitude superior ensuing from use team game we introduce formal mathematics for analyzing designing collectives. also suggest new utilities should outperform COIN certain kinds domains. accompanying work explain previous experimental results concerning superiority heuristics. make numerical predictions, some which then test. these two papers establish study collectives proper science, involving theory, explanation old experiments, prediction engineering insights.