"At the risk of sounding crazy, I have been looking at water vapor loops tonight and the pictures I am looking at unless they are way hours old, tell the story. The hurricane is going to go where the moisture is available."

There's nothing crazy about your observations at all. Here's my hunch in the context of the events of the last several days, and why. I'll start with the modeling. Clearly, CAT5 is beyond the range of available data. Any competent analyst/researcher that builds an empirical model provides ranges from sample data and precautions to users regarding extrapolations, i.e. projections beyond the ability of source data to estimate conditions, location, etc. But in the "real" world, by the time model output is dropped in the laps of (1) those that execute the programs, and (2) the end user, all precautions are gone or ignored. Mix in the politics and federal funds, and output mysteriously becomes biblical. Thus, each day the models screw up, folks go to bed, and then everyone starts a new day to see where the models point, all in an atmosphere of total effective amnesia regarding the previous day's failure. It's groundhog day all over again. I know there are a lot of sensitivities out there, but it is unreasonable to claim the models do better today than they did 20 years ago, and then use the statement to justify repeated gross error. So what. The models blew it back at Jamaica at best, and probably back in Grenada. End of story. One contributor last night posted an image of daily misses vs. Ivan's actual track. This was an interesting exercise that reveals persistent bias, but the approach suffers from, in a sense, serial correlation - errors each day that should be compounded, or propagated, and that are dependent on previous errors, are ignored. Once Ivan left the prediction "cone" many days earlier, meaning it effectively left the (presumed) 95% bounds for the predicted value(s) output by the models, the models had failed, the projected track left Ivan, and it's back to the drawing board for a new model. It no longer matters, from a modeling perspective, if Ivan now hangs a sharp right and does a 90 into Florida (just an example, folks). The models failed.

Granted, forecasting is extraordinarily complex and encumbered by sparse data. Folks in the way 2 or 3 days out need a feel for what to expect in the form of impending danger. Model runs will continue. But as Ivan virtually straightlined through the Caribbean, rather than shoot up the west Florida coast, posters here and presumably everywhere recognized the apparent land aversion. Even the special effort Ivan made over the last couple of days to avoid Cuba draws attention. As such, given this is a monster CAT5, there is not a reason in the world to believe any model was going to predict back at Grenada that this would be the path. There are simply too few observations (available data), and too many fundamental voids in theory, to expect any better. Nor do I recall in the casual surfing I've done during this circus that any forecaster called this path back at Grenada, though as a sheer matter of probability, one is out there somewhere. JB got close in his own way, but he bailed.

Thus, speculation about outlier hurricane behavior is by no means out of place. Nobody else has it figured out, that's for sure. Perhaps new thinking is required, and this is how the model revamping process starts. For example, maybe the apparent land aversion is not a land aversion at all. Maybe, for a hurricane of this strength, the rules change. Perhaps water vapor is the answer; here's my gut feeling. Perhaps a very large hurricane must be viewed not as just an intense cyclone, but as a body of (relatively) hot water looking for a release mechanism, thermodynamics in action, where guiding winds and low and high pressure systems are of much lesser importance. In this sense, Hurricane Caribbean ended last night, and with morning we say hello to Hurricane Gulf of Mexico. The cyclone itself is the energy release mechanism that has just by chance become available (where "by chance" means we do not have the understanding necessary to explain and accurately project the release mechanism's birth and transport). So while we watched Ivan's eye meander somewhat and the system appeared to avoid land, perhaps what was really happening is that the tremendous energy in the hot bath of the Caribbean, the core of what we are observing as an energy system, simply would not let go of the energy release mechanism. Similarly, perhaps water vapor has it's own role to play, since we are dealing with highly complex energy systems.

I grant these ideas can and will be easily criticized for being trivial, simplistic, ignorant and practically redundant. But nobody has gotten Ivan right yet. Importantly, we must question the wisdom of providing 5-day incorrect misleading error-laden forecasts, particularly for an outlying event for which there is little precedence and at best sparse data. Surely people died in accidents of various sorts (cars, heart attacks, fights, emotional standoffs and domestic disputes) as they fled or boarded-up under incorrect pretenses. We will not hear about them. To state, as some have, that it is better to be safe than sorry, is to succumb to the precautionary principle. Should we live lives of total paranoia? Maybe Long Island should be evacuated, just in case. Heck, we might as well evacuate Colorado, just in case. Just how safe should we be, and how wrong should models be before we finally draw a line and say wait - there must be a better way. After all, it's our taxes that are paying for the models, and taxpayers' lives that are endangered.

Sure, hurricanes are complex circulations that are tossing moist and dry air all over the place. But maybe there is something to your water vapor idea...