It's based on what we already know about previous quakes. The energy released is proportional to the surface of the fault plane that moves. All previous quakes in the M 9 range have had really long fault lengths, and have been suduction quakes. There are (fortunately) only a fairly small number of locations on the planet that can generate an M9. South of Java and Sumatra, Kamchatka, Alaska, off Washington and Oregon, and off South America. Things get complicated near Japan, I don't think you can quite reach M9 right near Japan because things are broken up into shorter faults. The 1755 Lisbon earthquake off Portugal is guestimated at M 8.7. There doesn't seem to be enough fault there to generate a 9. But they're actually not exactly sure where that quake was.
OK..thanks for that explanation..I understand the data,a nd the theory being cited..but it's NOT a proven, right..it's a theory, extrapolated from data, which is both not precise, and scarce..IOW, we could be infor a surprise someday, SOMWHERE. Again, I'm NOT arguing with you, per se..heck..I don't know enough about it to argue with myself..(G) but the analysis you described is like the old joke of blindfolded people describing an elephant by touching it..