NOAA's actually pretty accurate at depicting the number of named storms per season. 2005 was the one season they were really off on.
Other than that they're mostly within the range they predict.
2004 - (12 to 15) - 15 named storms
2005 - (12 to 15) - 28 named storms
2006 - (13 to 16) - 10 named storms
2007 - (13 to 17) - 15 named storms
2008 - (14 to 18) - 16 named storms
So you are saying that forecasting, even with computer models, if more accurate ... say 6 months out, than say ... what the average temperature of the planet will be in 100 years?