This looks pretty cool, but by default it gives Obama a 100.00% chance of winning? Something seems off about that. With the more conservative settings that you list for stddev and sample size, it gives more reasonable results. Why didn't you pick those as default?
Interestingly, it's only been giving 100.00% in the last few days as the poll margins widen. Actually, manually setting the standard deviation seems arbitrary to me. It's useful to vary the parameters to experiment and get a feel for how the statistics work. But it seems to me that the intrinsic statistics really are telling us something about the current state of the race and the amount of certainty inferable from the polls and electoral college combinations. Finally, I note that FiveThirtyEight's Now-cast probability is at 98.0% today, so 100% doesn't feel too off to me given the simplicity of this simulation's statistical model.