I too have noticed the westward shift of the numerical models on Irene. The BAM series now take the storm toward the east coast of Florida while the global models turn her NW toward Bermuda. What's interesting though, is that both the UKMET and NOGAPS stall the storm near Bermuda and then move her due west toward the Carolinas on the last day of their run - with both models showing a building ridge north of storm that turns her west. In addition, GFS, GFDL, and the CMC never intialize a storm & pretty much keep it an open wave. Lot's of uncertainty still in the final track and intensity. I'm not ready to write her off yet as a fish-spinner.

http://flhurricane.com/modelanimator.php?storm=9&Year=2005