the idea of AMR (accelerating moment release) as espoused by Sammis, Bowman, and sometimes Sornette has built in a scale several times the mainshock rupture length. Thus, an area 100s of km across would spawn an accelerating pace of earthquakes before an M8, somehow set before the mainshock occurs. Not saying I believe it, but it is not impossible, either. KB also has similar concepts in his methodologies.