Follow the Fast: A Simple Algorithm for Extracting Wisdom from Crowds

Abstract

Crowdsourcing a solution to an open-ended question often results in a wide range of answers. A challenge then is to identify the correct or most accurate answer. We propose a simple strategy: select the answer of the person who responded fastest. Using answers of TV game show contestants to quiz questions and diagnoses of physicians assessing a series of medical cases, we find that this algorithm performs better than random selection and following the slowest individual, but worse than following the most confident one. The method also outperforms following the best performing individual on other questions, unless this expert is selected from a large crowd and their skill can be reliably assessed. The accuracy of the follow-the-fast algorithm improves with crowd size. These findings show that following the fast can be an effective method for extracting wisdom from crowds, using a cue that is often readily available.