Crowds within crowds are wiser still
Argentinian research finds small-group deliberation much more accurate than large-group guesses. Stephen Fleischfresser reports.
Famously discovered by Charles Darwin’s cousin, Francis Galton, the “wisdom of crowds” is the phenomenon in which the aggregation of estimates from a large number of people is far more accurate than the estimates of the individuals that make up the whole. Recent research has shown once again that the wisdom of the crowd comes out on top against other forms of aggregated estimates. Understanding the idea is vital to comprehending the way humans go about the collective decision making that is so central to political life.
It’s a fascinating phenomenon, but one prone to distortion, particularly if individuals in the crowd begin to influence each other. Social influence can make the crowd less wise because it leads to “herding”, where the mob converges on a consensus at the expense of accuracy. Previous research suggests that individuals within the group should remain independent to maintain the wisdom-of-crowds effect.
A team of international researchers led by Joaquin Navajas from Universidad Torcuato Di Tella, in Buenos Aires, Argentina, has now published experimental results that go against this thinking.
The team conducted an experiment at a TEDx event in Buenos Aires in 2015, as well as a follow-up in the laboratory, both questioning the idea that members of the crowd should act independently of each other. Instead the team asked, “can crowds be any wiser if they debated their choices?”
In the first experiment, 5180 people were asked to answer questions based on general knowledge that involved a numerical estimate, such as, "What is the height in metres of the Eiffel Tower?" Respondents first answered the questions individually, then collectively through discussion in groups of five.
The results were astonishing.
Aggregating the estimates of the smaller groups was more accurate than aggregating individual estimates. The team’s analysis concentrated on 280 groups of five (thus, 1400 people), and found that “the average of four collective estimates was even more accurate (by a 49.2% reduction in error) than the average of the 1400 initial individual estimates.”
A “crowd of crowds” seems to be wisest of all.
Why? Well, numerous prior studies have suggested that, as Navajas and colleagues describe it, a “fundamental condition to elicit the wisdom-of-crowds effect is the diversity of opinions.” The team’s experiment seems to indicate that while this social interaction decreased variance in estimates within the group of five, it actually increased variance between the groups. This promotes an even greater diversity of opinion and results in a more pronounced wisdom-of-crowds effect.
In the follow-up experiment, the team discovered that participants in the small groups came to a consensus via “shared arguments” and collective reasoning. This kind of deliberative “crowd of crowds” was more accurate again. The researchers found that “the average of four group estimates collected in experiment 2 was significantly more accurate than the average of all 5180 initial estimates collected from the crowd”.
Taken together, this demonstrates that the crowd gets wiser “by simple face-to-face discussion within groups coupled with between-group sampling,” a result that “supports political theories postulating that authentic deliberation, and not simply voting, can lead to better democratic decisions.”
“It is difficult to overstate the importance of these findings,” say the authors, “as they call for re-thinking the importance of the deliberation structure in joint decision-making processes.”