Dennis...I'm at library to go online and have a lot of email to deal with...I don't have much information on this, and since I am not at my home office, can't dig out the couple articles that I saved on this subduction zone. But, not looking at a map, the subduction zone is arcuate coming NW from this quake along Sumatra to the 2004 and 2005 quakes. There, the plate motion is more right-oblique to the boundary, and the right-lateral component is partitioned onto the Great Sumatra fault. Down by Java, the plate motion is closer to pure subduction. I suspect you are correct that there is some kind of significant velocity or geometrical discontinuity where the Great Sumatra fault splits off, which could serve as a rupture boundary. We have discussed here in the past the 1833 and 1860something quakes south of the equator that Kerry Sieh's group and web page have studied...these have accumulated enough strain since then for a great quake, which could be any time, or 100 years, but I'd guess one or both of those patches would fail closer to today than to 100 years from now...
Chris in R.I.