Ground Hog Day 2012: Revise That Equation

I have the distinct impression that I've written this post before. Several times. You may feel that you've read it before. Over and over again. It's only 8 am ET here, but already (thank you Internet) the results and prognostications are in:


Date Prediction Groundhog Location
2012 Early spring[ Wiarton Willie Wiarton, Ontario
2012 Early spring[23] General Beauregard Lee Lilburn, Georgia
2012 6 more weeks of winter[24] Malverne Mel Malverne, New York
2012 Early spring[25] Holtsville Hal Holtsville, New York
2012 Early spring[26] Buckeye Chuck Marion, Ohio
2012 Early spring[27] Staten Island Chuck Staten Island (New York City)
2012 Early spring[28] Shubenacadie Sam Shubenacadie, Nova Scotia
2012 6 more weeks of winter Punxsutawney Phil Punxsutawney, Pennsylvania

I knew the "rule": if the GH sees its shadow, there are 6 weeks more of winter. I thought incorrectly that if the GH didn't see its shadow, there are 6 more weeks of winter anyway. But I have now been disabused of this idea: if the GH does NOT see its shadow, the equation tells us "Spring will be early". How early? Nobody says. So the current, traditional theorem is this:

f(GH) = sees shadow = 6 more weeks of Winter
f(GH) =does not see shadow = early Spring

The Wiki tells us about these predictions:

Groundhog Day proponents state that the rodents' forecasts are accurate 75% to 90% of the time. A Canadian study for 13 cities in the past 30 to 40 years puts the success rate level at 37%. Also, the National Climatic Data Center reportedly has stated that the overall prediction accuracy rate is around 39%.

Did you see that? According to the Canadian Study and the National Climatic Data Center, we've got it exactly backwards. Completely backwards. Taking a contrarian view of the current theorem,

f(GH) = t, where t = duration of winter

,

nets you an astonishing 63% or 61% accuracy. Way better than a coin toss. More powerful than a speeding locomotive. Able to leap tall buildings at a single bound. And over time almost (almost is a relative term) accurate 2 out of 3 years.

The question, given all of this, is why we persist in the current formula. The current equation is clearly incorrect. And so a simple proposal from here into the future. Can we change the equation to this:

f(GH) = -t

I'm not expecting a vast cultural uprising about this, I'm not even expecting anyone to notice. We just keep doing it over and over again.

Happy GH day!