`That looks weird' - evaluating citizen scientists' ability to detect unusual features in ATLAS images of LHC collisions

Using data from the HiggsHunters.org project we investigate the ability of non-expert citizen scientists to identify long-lived particles, and other unusual features, in images of LHC collisions recorded by the ATLAS experiment. More than 32,000 volunteers from 179 countries participated, classifying 1,200,000 features of interest on about 39,000 distinct images. We find that the non-expert volunteers are capable of identifying the decays of long-lived particles with an efficiency and fake-rate comparable to that of the ATLAS algorithms. Volunteers also picked out events with unexpected features, including what appeared to be an event containing a jet of muons.