Electromagnetism and time-asymmetry

It is a commonplace to note that in a world governed by special or general relativity, an observer has access only to data within her past lightcone (if that). The significance of this for prediction, and thus for confirmation, does not however seem to have been appreciated. In this paper I show that what we regard as our most well-confirmed relativistic theory, Maxwell's theory of electromagnetism, is not at all well-confirmed in the absence of an additional assumption, the assumption that all fields have sources in their past. I conclude that we have reason to believe that there is a lawlike time-asymmetry in the world.