The 'parent' model of the GFDL (the GFS) did not change that much from its last run, so the variation in GFDL output from run to run indicates that small differences could have a huge effect on the eventual track. The overall predictability of this scenario seems marginal at best, so there will likely be some uncertainty to deal with for awhile. Whether or not Wilma actually makes landfall in the Yucutan or not would make a huge difference on the eventual intensity as well, so the intensity forecast is even more uncertain, since it hinges on an uncertain forecast track, not to mention ERCs.