On 28 June 2022, Émile Servan-Schreiber reshared Arjun Vir Singh’s account of a presentation at Arthur D. Little’s partners event in Faro, Portugal. Singh described a talk by Servan-Schreiber from the UM6P School of Collective Intelligence and argued that diversity could act as fuel for a group’s intelligence.
The attached collage records the presentation title, Collective Intelligence: How Diversity Makes Us Smart, against the event’s Together Again, Faro 2022 branding. Other photographs show a slide presenting voting, averaging and betting under the heading of objective aggregation. The material connects the composition of a group with the method used to combine its contributions.
An attendee’s conclusion, with qualifications
Singh suggested that choosing diversity even at a cost to meritocracy could sometimes be right for an organization. That is his interpretation of the presentation. It raises a selection question for managers: should a group consist only of the strongest individual performers, or can someone with a different perspective add information the others lack?
A useful mathematical context is the diversity prediction theorem. For numerical estimates combined by a simple average, the squared error of their mean equals their average squared error minus the average squared distance of estimates from that mean. At a given level of individual error, different estimates can make the average more accurate. Selecting a different group, however, can also change individual error; diversity alone does not settle which group will perform better.
In a University of Michigan talk, Scott Page explained that crowd performance depends on both individual ability and differences in how people see a problem. He warned that copying others can reduce useful diversity. His explanation connects the mathematics of averaging with a practical concern: contributions must retain differences if those differences are to help the combined estimate.
For forecasting, the practical issue is how to combine competence with contributions that add information. Servan-Schreiber’s later discussion of expertise and diversity develops that question. The Faro presentation brought group composition and aggregation into the same conversation for a consulting audience. The distinction is useful: changing who contributes and changing how their answers are combined are separate choices, each affecting the resulting judgement.
