If i understand it correctly, there seems to be a big problem in the design of the evaluation. Namely the assumption that you only have to identify expert voices. That sounds great for simple and isolated problems, but those are not the ones we are facing today. We live in a complex system were every expert has a very narrow perspective and problems are too complex for any one individual to grasp all relevant aspects of.

The difficulty is not finding the one person that is competent enough to fix the problem for you but finding someone that is able and willing to create a process/environment in which the relevant perspectives/expert opinions are collected and a cooperative solution is found (to address the relevant perspectives on the problem).

By voting or delegating for a solution finding process instead the selection of one solution you overcome the "everyone needs to be an expert and inertia / disinterest sets in" problem identified here.

tldr: I believe the study tests for compatibility to a non-functioning legacy/simplified approach to decision making.

----------------------

edit: From the paper

>The first experiment was designed for the lab and follows the theory very closely. We study groups of either 5 voters (of which 1 is an expert) or 15 voters (of which 3 are experts). We observe the frequency of delegation and the fraction of group decisions that yield the correct outcome.

>Our second experiment then is meant to capture a voting environment where voters have “some sense” of how well-informed they are and how likely they are to be correct, and similarly of how likely experts are to be correct, but such sense is vague and instinctive.

The assumption that you have to simply identify the person that has the correct solution is fundamentally flawed.