Collective intelligence: a term we often see and hear these days. Scientist do not currently understand whether the outcomes of predictions that were based on collective experience can really be considered “intelligent” or “collectively intelligent.” Plenty of examples point to the fact that groups of people have had considerable success making predictions, by using calculations that were based on their individual predictions. But still no-one knows when groups of people can be relied on to make predictions, or not. In July of this year, the Institute Organization Management (IOM) at Steinbeis University Berlin embarked on empirical research to investigate the phenomenon of collective intelligence.
Code-named KnowledgeCloud, the project aims to monitor various predictions made over two years in all areas of media and society. Predictions will be made by asking as many people as possible. The subjects do not have to be experts in a field; after all, the underlying aim of the study is to examine whether collective intelligence is a product of many people thinking together from a variety of different backgrounds.
The first round of KnowledgeCloud predictions ended in August. It looked at the market share of different smartphone operating systems in the first quarter of 2012.
By the summer of 2012, it will be possible to see how good the “many” (170 people) were at making a prediction and if there are any first indications that collective intelligence exists, is growing or is non-existent. The second prediction round starts in October 2011. People can still take part by submitting predictions online to the following: What will the turnover of eBooks in fiction be in Germany for the whole of 2012?
Based on 2010 sales, which were Euro 20 million, and predicted sales of Euro 68 million in 2011, the German Publishers and Booksellers Association estimates that the 2012 sales of eBooks in 2012 will be Euro 136 million. The IOM is hoping that the KnowledgeCloud project will identify whether a collective prediction will be more accurate.
Researchers at the IOM will only be looking at short and medium-term predictions for their experiments, focusing on topics for which they have data or, where applicable, predictions from other sources. Once the two-year study is complete and many such experiments have taken place, it should be possible to spot patterns in the performance of the “collective intelligence” for these kinds of predictions. To work out the patterns, it is important that as well making a prediction, research participants must state how familiar they are with the field they are making the prediction about. This is so the scientists can analyze the extent to which being familiar with a topic improves or worsens performance.
For example, for the eBooks prediction, respondents were asked if they have ever bought an eBook or read one. This allows the IOM to evaluate whether predictions made by the volunteers with eBook experience are better, the same or worse than the respondents without such experience.
Based on this method, the KnowledgeCloud will generate a continual stream of different pointers over the next couple of years: