This is from a football column I read throughout the season, but at the end of the season the guy gives a recap of bad predictions. This was included in his column a couple of weeks ago (He's not just a sports writer, he is also "a contributing editor for The New Republic, The Atlantic Monthly and The Washington Monthly, and a visiting fellow at the Brookings Institution.")
http://sports.espn.go.com/espn/page2/story?page=easterbrook/070213The year of Katrina and Rita, 2005, obviously was awful for Atlantic cyclones, with a record 15 hurricanes. Both the National Oceanographic and Atmospheric Administration, and media-favorite hurricane forecaster William Gray of Colorado State University, predicted 2006 would be bad, too. In December 2005, Gray predicted for 2006 nine hurricanes, five of them intense; there was an 81 percent chance a major hurricane would strike land in the United States in 2006, Gray and CSU said with ridiculous pseudo-precision. In May 2006, NOAA forecast eight to 10 hurricanes, six of them intense. The 2006 hurricane season would be "hyperactive," NOAA declared: "The main uncertainty is not whether the season will be above normal but how much above normal it will be." Actual: In 2006 there were five Atlantic hurricanes, two of them intense -- pretty much smack on the 20th century average for the Atlantic basin. None made landfall in the United States.
What's amusing is not so much how far off the predictions were but how the forecasters tried to weasel out of their projections. Late in August, after most of the summer already had passed with calm seas, NOAA switched to forecasting only seven to nine hurricanes. Gray and CSU switched to forecasting seven hurricanes, three of them intense. That is -- now that we know there haven't been a huge number of hurricanes, we are predicting there won't be a huge number of hurricanes! In August, CSU lowered its prediction of a major landfalling hurricane from the ridiculous pseudo-precise 81 percent to an equally ridiculous pseudo-precise 73 percent.
In September, Gray and Colorado State changed predictions a second time, "forecasting" five hurricanes. That is, once it was nearly certain 2006 would end as an average year for hurricanes, the hurricane experts predicted an average year. Unable to resist its utterly meaningless pseudo-precise forecasts -- issued, perhaps, to impress science-illiterate journalists -- Colorado State said in September there remained a 59 percent chance a hurricane will affect the U.S. coast that month, plus a 14 percent chance a hurricane would strike the United States in October. Needless to say, none struck in either month. Note to the experts at Colorado State -- when you multiply a bunch of estimates, do not treat the product as precise.
Gray and CSU might as well have claimed there was a 59.263452096489244 percent chance of a landfalling hurricane. Hilariously, the Associated Press story about the September backtrack by Gray and Colorado State declared, "Previously, Gray had predicted seven hurricanes." Wait a minute, at first Gray predicted nine hurricanes. Then after he knew there would not be a lot of hurricanes in 2006, he started claiming he had predicted seven. After the hurricane season was nearly over, he started claiming he predicted only five. When media hurricane concern ramps up around Memorial Day, bet you anything Gray tells gullible reporters, "I was the one who predicted there would be five hurricanes in 2006."