Using Space Warps To Find Gravitational Lenses

Gravitational lenses are very useful astrophysical tools, providing direct mass measurements of galaxies and clusters, a magnified view of the distant universe, and a probe of both dark matter and dark energy in the universe. Key to extending the science made possible by gravitational lenses is the discovery of more systems in wide field imaging surveys, but their complex nature makes finding lenses a difficult process to automate. Space Warps is a citizen science project that enables new gravitational lenses to be discovered by volunteers inspecting images displayed to them via a custom-built website. Over 60,000 volunteers have taken part in the Space Warps project to date, and have found dozens of lens candidates. The classifications made by the volunteers on both the training and test images are interpreted by a bureau of software agents, in an implementation of a probabilistic model with over half a million parameters: we infer the skill perceived by each agent as an expected information contribution (in bits), and use this to assess the properties of the crowd during the assignment of lens candidate probability to each object. I will give a guided tour of the project, and point out some general lessons we can draw from it.