Optimal collective decision making: consensus, accuracy and the effects of limited access to information.

"Knowledge is power"-holds the popular proverb, because knowledge and information is indeed one of the cornerstones of effective decision making, a requisite all living beings face continually. In fact, effective decision making is a matter of life and death, for individuals and groups alike. Furthermore, in case of group decisions, consensus is also often desirable. This latter one has been studied extensively by means of formal (mathematical) tools (in the field of opinion dynamics), while the first requirement, the process of yielding accurate information has been largely neglected, at least so far. In the present paper we study the optimal structure of groups which are embedded into an external, observable environment for (i) reaching consensus (ii) having well-informed members, and (iii) for those cases when both aspects are equally important. The groups are characterised by their communication networks and individual properties. We find that the group structures fundamentally differ from each other since having well-informed members requires highly specialised individuals embedded into a structured communication network, while consensus is promoted by non-hierarchical networks in which individuals participate equally. We also find that-contrary to intuition-high access to information calls forth hierarchy, and that suggestibility promotes accuracy, not consensus.