Cost of Consensus: Overconfidence in Group Decisions
There is a curious confidence in the power of consensus in bureaucracies and institutions -- including institutions of government and inter-government which increasingly set limits on the future of humans. And yet consensual decision-making is rarely put to the test, to assess its weaknesses and how it might be radically improved.
We demonstrate that the very process of making a judgment collaboratively rather than individually also contributes to such myopic underweighting of external viewpoints. Dyad members exposed to numerical judgments made by peers gave significantly less weight to those judgments than did individuals working alone. This difference in willingness to use peer input was mediated by the greater confidence that the dyad members reported in the accuracy of their own estimates. Furthermore, dyads were no better at judging the relative accuracy of their own estimates and the advisor’s estimates than individuals were. Our analyses demonstrate that, relative to individuals, dyads suffered an accuracy cost. Specifically, if dyad members had given as much weight to peer input as individuals working alone did, then their revised estimates would have been significantly more accurate. _Minson and Mueller 2012 AbstractThe researchers discovered that the larger the number of the collaborative consensus group, the greater the danger of this overconfident refusal to entertain outside input.
The unwillingness of collaborative "insider" groups to entertain outside input is particularly dangerous when the results of the decisions made by such groups could lead to poverty and oppression for large numbers of people. One ready example of such a dangerous consensus is the IPCC collaborative of "ClimateGate" fame. An honest and thorough assessment of insider emails between IPCC insiders reveals an internal climate of hubris, overconfidence, and a willingness to suppress contradictory evidence and to ruin the careers of anyone willing to espouse other points of view.
This willingness of group members to subsume themselves within the identity and purpose of the group -- this decadent conformity of consensus with its concomitant blockheaded overconfidence -- becomes a more serious problem as more and more important decisions are being made by bureaucratic committees and collaboratives.
Individuals are not, of course, immune from this group consensus fallibility. In fact, within a single brain, decisions are made by competition between different brain-wide neuronal assemblies, which battle it out until a winner emerges. That type of decision making is a consensus of sorts, a trial by combat that takes place on a subconscious level constantly, as choices and decisions are made on a moment by moment basis.
Over time, overconfidence is possible on an individual basis, if enough subconscious decisions and choices result in success. Daniel Kahneman has studied this phenomenon of intuitive overconfidence:
"What you find is a great deal of confidence in the presence of very poor accuracy," Kahneman explained. "So the confidence people have is not a good indication of how accurate they are."It is not necessarily a question of whether individuals or groups are better at problem solving or decision making, although false confidence tends to grow exponentially as the numbers in the group grow. The important thing is for both individuals and groups -- no matter what the purported level of expertise -- to be relatively open to looking at alternative approaches and theories.
Overconfidence is accentuated by the failure of people to, in general, learn from their mistakes. "When something happens that a person has not anticipated, ... they remain convinced that what they had predicted, although it didn't happen, almost happened," he said. The overconfidence is then propagated while the accuracy remains the same, and the cycle begins again. _APS
The modern curse of conformity in general, is a much tougher nut to crack. We will be dealing with that question both here and at Al Fin the Next Level.