A storm may wobble in a loop on its own accord when steering currents are weak, or may traverse a loop when the steering currents are forecasted to change. For instance, say Jeanne begins to travel north and then northeast as a trough picks it up. But, the trough then moves away without completely picking up Jeanne and a ridge builds in over the storm. This would tend to shift the storm back west, but may result in a loop to do so -- due to the prior movement and speed of the storm (think momentum), it can't simply stop on a dime and turn west.

And really, there's no one out there who can definitively say this storm will not affect Florida -- or any landmass out there, for that matter.

As an aside, when the models offer wildly diverging options -- think of the tracks spreading out in a pattern like a spider -- the course of least regret is to forecast a slow motion through the forecast period. This pattern usually happens when none (or few) of the models are getting a good handle on the storm -- and with a weak storm trying to reform like Jeanne, this certainly qualifies.