Michael...my reading of your post and what I have heard elsewhere both make me think the usage is both spatial and temporal: the hypothesis depends on there being spatially-defined fault segments. But, if there is a 200 km spatial gap between, say, 2 Great earthquakes that occurred 10 years ago, but there was one on the intervening gap 30 years ago, and it slipped 5 m on a subduction zone accumulating strain at 50 mm/yr, it would not be "due" and so not identified as a gap on that basis...
Chris