According to a widely accepted view, the current ICT trend will enhance the wisdom of crowds and challenge the role of experts. For instance, projects like Wikipedia already outperform traditional knowledge systems based on specialists’ effort. Still, this scenario encounters some residual resistance. Recently, the Pachube staff received a comment on its blog asking: how can you ensure the accuracy of the feeds gathered on your webpage? Pachube replied:
The beauty about crowdsourcing is that the dataset draws strength from its heterogeneity,
meaning that many signals of uncertain or low quality can result in an aggregated signal of high usefulness: uncalibrated outliers would be easily spotted, ups and downs would be observed promptly. And you need a social media such as Pachube to reach this result: the more people send sensors data to the website and see their point on the map, the more people are encouraged to do so, in order to create a more impacting picture of the monitored phenomenon.
A recent paper, published on Pnas, reveals that the wisdom of the crowd can be lost if the crowd interacts and exchange information. The ETH team composed by Jan Lorenz, Heiko Rauhut, Frank Schweitzer and Dirk Helbing asked volunteers questions like
How many murders were officially registered in Switzerland in 2006?
Researchers shown that while the crowd is able to guess a measurement with high accuracy if individuals are isolated, their mutual influence moves outliers toward the average estimates, but does not change substantially the global estimate.
The paper has triggered a great discussion within the scientific community. However, its findings are not surprising. Any social network regular user knows that social media can be a powerful channel to spread inconvenient truths but also false beliefs, since imitation is a popular strategy among network users. A social media, however, does not only provide tools to put people in contact – a rather trivial task nowadays. It provides also – at least in the best cases – a filter for relevant information and reliable sources, with many possible mechanisms at work. If such filter does not exist or is not efficient enough, no crowd will be wiser than a herd of sheep.