The Role of Task Difficulty in Collective Intelligence
A new study has explored whether crowd intelligence works equally as well regardless of the difficulty of the task being attempted. The researchers devised a number of tasks for the crowd to perform, with the complexity ranging from answering “if all poor people in the world gave you $1 each, how much sustainable monthly income could you derive from the resulting amount?” at one end, to “what is the high temperature in Seoul today?” at the other (Seoul was the home town of the participants).
The 500 participants had to answer 8 of these tasks, with no prior revision allowed beforehand, and methods undertaken to offset the possibility that anyone could look up the answers online, with the data produced from the tasks suggesting none of the participants did.
The results showed that the power of the crowd was most prominent when the questions posed were in the mid-range of difficulty, or in other words, when the task required access to a wide range of specialized information. These kind of situations saw the crowd do significantly better than the individual.
By contrast, when the tasks were very easy, both individuals and the crowd performed well, so there was little to separate them. That is arguably to be expected, but of more interest was the performance of the crowd in the hardest task.
Here it emerged that the individuals did better than the crowd, with the data suggesting that participants were generally so clueless as to the answer that they resorted to random guesses, thus providing a very wide spread of answers that even aggregation could not help to converge on the true value.
So, it seems that if you want to truly capitalize on crowd intelligence that you need to ensure the task is neither too difficult, nor too easy. That isn’t to say that hard tasks cannot be tackled of course, but merely that such tasks need to be managed and reduced down to moderate levels by the crowd themselves.
Other Posts by Adi Gaskell
Social Media Today