Glen, it would be next to impossible for the North Anatolia fault (NAF) to have
a M9earthquake, and unlikely for it to have a M8; A M7.5 would be a lot easier. That is because Magnitude is related to the fault area that breaks, and to the slip (and to the stress drop). The NAF is a right-lateral fault, and it is steep. The part near Istanbul that can break in an earthquake would break only to about 20 km depth (possibly a little more in the west, a little less near Istanbul). So, the downdip width might only be around 20 km, while for a relatively flat subduction zone, it could be 100 km or 200 km (someone can check how wide the M9s were: Sumatra was a little narrow compared to Alaska and Chile, I think). The other factor on fault area is the alongfault length: Sumatra was about 1400 km, Chile was about 1000 km, and Alaska was 600 km. Because most of the NAF broke during the 20th century, it is highly unlikely to fire off a quake along its wholelength: the most likely length would be the distance between the Aug 17 1999 quakes west rupture in Izmit Gulf of east Marmara Sea to wherever offshore the 1912 Ganos quake slip became minor or stopped in western Marmara Sea.

The slip in a strike slip quake is rarely more than 10 m, and over a long rupture might average half that. Sumatra was locally 20 m, and I've heard very large numbers for slip on Chile.
The same logic applies to the San Andreas fault: it cannot have a M9, but a M8 is more possible on it than on the NAF, because one cannot rule out the "wall to wall" quake from SE of Parkfield down to the Salton Sea (although unlikely).
I would not 100% rule out a M8 on the NAF because there are other strands to the south that are probably active. Except for Parkfield, the wall to wall scenario would affect parts that have not broken, from north to south, since 1857, 1812, and probably 1700.

Hey...I wonder if the southern San Andreas quake about 300 years ago could have occured at the sametime as the last Cascadia quake, which was also 300 years ago...in 1700.
Chris