Your statement concerning verification is "right on". That is why the monthly forecasts are being catalogued on separate pages this year. That way, each forecast can be verified after the season is complete. If you will notice, a major change is to include the previous month's forecast links on each succeeding month.
There is an ongoing process for "re-loading" the graphics from the past seasons. The major hurdle is that there was such a hurried move to new servers and hosts last year that a considerable number of the early pages are still being recovered.
As to "revisionism", you are correct as well. The forecasts have been made public since 2002. Every attempt has been made to make sure that they were "viewable" as early as possible so as to avoid just that situation.
There is currently a project underway to "back verify" the model that is being used. Those results will be posted as soon as they are available.
As to the "infancy" of the process, again you hit the nail on the head. This is te 4th year of the specific location for landfall forecasts. Undoubtly, 2004 was a unique situation, 100% is well beyond reasonable expectations. The prior year's of 80% and 70% were actually better than was expected for a starting point. It does show though that the model, which has been "tweaked" this year again, has some merit. It just isn't perfect yet. But, there is every hope that it can continue to be improved enough in the coming years that 85% to 90% will become the norm.
Remember, the forecast is NOT for any specific strom, but for impact by A storm.
