The real probabilities of "wall to wall" ruptures are bound to be small, and strike me as a distraction from the overwhelmingly dominant hazard from M7 to M8s.

Figuring out the scenarios is complicated by the apparent scaling of slip with rupture length, although one might expect slip to scale only to rupture width. For example, if a 400 km rupture has up to 10 m of slip, does a 1000km rupture have 20+ m?

As for the expected high-frequency motions from big events, that is a real can of worms. Some suggest that slip more than a meter is smoother than smaller slip, and I've heard Somerville argue that bigger magnitude can mean smaller hi-freq motions in some cases. The attenuation structure matters, and is much more poorly constrained than the velocity structure. And we don't know the velocity structure very well at short wavelength either.