I'll speculate that it may have been hard to trigger the 1811-1812 quakes by water. Some think that all the seismicity since then is aftershocks (I read a recent publication on this). Aftershocks last far longer than in, say, the San Andreas system because the blocks are not moving past each other (tectonic stresses are accumulating extremely slowly). So, stresses redistributed by main shocks, and maybe easier to trigger small aftershocks. Now I'm starting to think that an earlier argument of Brian's may be relevant. Repeated flooding of Mississippi would mean that load-induced triggering may have already happened, if any.

OK: Brian's math looks fine, is interesting. 10 km column of granite is 2.65x more mass than 10 km of water. But, we know that very small changes in shear stress at a fault even 10 km down can trigger quake. You have to differentiate total stress (which is what Roger was talking about) from differential or deviatoric stress (I mix the last 2 up and don't feel like Googling this). Total stress is 1000s of bars at 10 km. The stress drop during quakes is usually around 100 bars, maybe less. Stress changes of a couple of bars can trigger quakes. And, pore pressure increase in a fault directly weakens the fault.
Chris