A Hurricane Complaint

What's with the National Hurricane Center (NHC) and its lame conservative hurricane-path forecasts? They seem to think the best approach is to poll all the models and choose the result that's most in the middle. Model results diverge from each other not for any stochastic reason, however, but because some models are better than others. A better approach is to choose just the best models and use their results. Determining which models are better for particular applications takes a lot of experience, however.

NOGAPS did really well forecasting the paths of Ivan and Jeanne, but as good as it did, its results were well outside the consensus approach. It took an unconscionable amount of time for NHC, kicking and screaming, to finally recognize the other model results were crap. But it was obvious the results were crap at the time - I could see it, and any skilled hurricane meteorologist should have been able to see it as well. So, what gives?