Something unusual happened with Dean. Due in part, or in full, to his near due west movement.
The SFWMD plots on the main page were easy to follow. The Extrapolated or XTRP line was close to the actual track prior to the first landfall.
http://image2.flhurricane.com/images/2007/storm4/plot18082007-2041.gif

GFDL did good at the 5 day prediction, but kept flip flopping between the Yucatan and Grand Isle,LA for a few days.
http://image2.flhurricane.com/images/2007/storm4/plot15082007-1402.gif

Here's the plot from the 16th, actually 10PM EDT on the 15th.
http://image2.flhurricane.com/images/2007/storm4/plot16082007-0156.gif

note:all the above links are time sensitive for Storm 04~danielw

Here's a table breakdown of what the various modle performances were.
http://hurricane.methaz.org/tracking/ops/AL042007/AL042007_perf.html