Of course you realize that "broken" sections of fault have a very heterogeneous distribution of slip. The Landers fault, the 1906 San Francisco rupture, the 2004 Sumatra rupture - all alternate patches with much slip with patches with little slip - so why can't the patches with little slip go tomorrow? Like the patch between the Dec 2004 and the Sept 2007 Sumatra rupture? Or even smaller patches directly under cities?

It is hard to say any fault has completely broken and is thus safe, even if one buys the nearly complete stress drop dogma.