Bruno Latour has taught that the success or failure of a science project has to do with scientists’ ability to enrol both human and non-human actors, and keep them engaged within a network. Yet the discipline of seismology has been unable to demonstrate a level of control over its study object that even meteorology, which initially shared its predicament, increasingly displays. One manifestation of this uneasy alliance between scientists and earthquakes are shifts in disciplinary branding. Branding, in the sense I’m using it here, is the creation of an identity or image one regards as an asset. At some times and places, the perception that seismology was paving a road to earthquake prediction has been an asset to its practitioners, while at others, prediction has been considered a liability to seismology’s status among the sciences. This has resulted in the creation of different brands of seismology, tied to different objects, projects, and even geographies, and also varying over time and opportunity. I describe them broadly as predictive and anti-predictive, allowing that there are areas of overlap, and that what counts as prediction has also been a moving target. While prediction of natural phenomena is considered a sine qua non in modern science, seismology demonstrates that a discipline can substitute other goals while maintaining a sense of its own progress, even relegating prediction itself to the realm of pseudoscience.