Login

Time on a large professional telescope is one of the scarcest resources in science. Astronomers wait months for a few hours, and every one of those hours is a running argument with the weather, the Moon and the atmosphere. Point at the wrong target at the wrong moment and the data comes back soft, washed out, or useless, with the next chance half a year away. A team has now handed that argument to a machine. Their system was trained not on the rules astronomers use but on what astronomers actually did, night after night, for years.