A large fraction of the images in Snapshot Serengeti are “blank”, i.e. they have no animals in them. This is because grass waving in front of the camera traps can cause them to fire. While doing some work on trying to remove the blank images in order to make the project more efficient, we stumbled upon an interesting phenomenon. It looked like the fewer blank images a volunteers saw, the fewer classifications they performed in total! This led to Zooniverse web scientist Alex Bowyer setting up an experiment to test this hypothesis. His very interesting and intriguing finding reads:
What we have shown in this experiment, is that reducing the percentage of “blank” images that a user sees in their session will reduce the length of time they spend providing voluntary classifications. This is very important for citizen science researchers to understand, since a common approach is to remove blanks in order to optimise for efficiency and not waste volunteers’ time. What we have shown is that optimising for user experience, not efficiency, is critical if each user’s contribution is to be maximised.
This continuing research is now going to look at other factors that might motivate volunteers to stay longer, such as variety and complexity. Read his full write-up of this experiment and its results at www.mico-project.eu/why-citizen-science-researchers-shouldnt-remove-mundane-images.