Improving the Performance of Collectives in Joint Problem Solving: Exploring the Impact of Task Difficulty, Collective Size, Intelligence Mechanisms and their Moderation through IT-enabled Task Shaping and Solution Aggregation

Description

Complex problems can often be better solved or only solved by a collective with knowledge or information drawn from a broad pool of expertise. For example, when SARS hit Hong Kong in 2003, the “clusters of illness were puzzling” (CDC, 2013) leaving the medical community confused. A turnaround came when people who weren’t medical experts were brought into the team. The resulting ability to identify the “super spreader” became a breakthrough in containing the disease. This ability of collectives to perform better than individual experts is by now well recognized (e.g., Surowiecki, 2005). Nevertheless, the results are equivocal, sometimes being better (lauded as wisdom of crowds), sometimes being worse (criticized as madness of the masses). Thus the phenomenon requires a further understanding of (1) why collectives outperform individuals, (2) which problems this strength is best applied to, (3) how this strength can be improved further, and (4) how it might be transferred to other domains. In short, we wish to find out why collective intelligence works and how to make it better, plus where it works and where else it could be more broadly applied. From a theoretical perspective we thus seek to “open the black box” of collective problem solving and determine the nonlinear effects of collective size, task difficulty (perceived and actual) and individual intelligence exploitation on collective performance. Given the affordance of information technology (IT) in remodeling task structure and aggregating individual intelligence, we attempt to further investigate the amplifying effects of task shaping and IT--?enabled collective aggregation mechanisms. We aim to also determine whether techniques of problem or solution structuring, whereby the problem is broken down into simpler tasks, or the solution space is narrowed down upfront, can help the collective perform better. Furthermore, we intend to explore the extendibility of the findings from collective problem solving to collective task work in domains such as open content creation (e.g., joint wiki editing). In addition to providing theoretical insights into the problem solving processes of collectives, we expect the research to have significant practice relevance, by giving guidance on how to improve collective solution finding in the field. The potential applications are manifold, from large problems such as improving Hong Kong’s public safety from disease, to very specific tasks such as predicting individual property prices.