A Criterion-Mining Method for Group Idea Selection -- Increasing Consensus with Minimal Loss of Efficiency

The fastest way for a group of experts to select raw ideas is in parallel. However, since each expert then no longer sees the large majority of the individual decisions, there is a danger that consensus about the result will be low. Ideally, the result of the parallel selection would be identical to the one that the group would have produced collaboratively. One cause of deviation of individual selections from the ideal case are hidden profiles: each expert works with their private mental model of the raw ideas and the selection criteria. Our hypothesis is that it is sufficient to build a shared mental model of the criterion in order to achieve consensus on the overall selection result: it is not necessary to discuss the ideas themselves. Our experimental results with a new selection method suggest that this is the case. In this manner, the motivation of the experts is maintained at little extra cost.