With clean water growing scarce in much of the world, and with shortages possible in 36 U.S. states in the next year, according to the General Accountability Office, the survey conducted for General Electric Co. found 66 percent of Americans feel positive about water re-use” -Reuters Article this week
Unbelievable, this 36 state prediction was alive. Wasn’t it dead? I wrote about this in 2007, but the press still uses this reference to hydrate their water articles. This zombie prediction may never die.
Don’t believe me? I found these online in ten minutes:
- “Experts do agree: Demand is greater than supply. And 36 states face water shortages in the next three years.” America’s Dwindling Water Supply, CBS NEWS 2010
- “At least 36 states expect to face water shortages within the next five years…” Drought Parches Much of the U.S., CNN 2009
- “The government projects that at least 36 states will face water shortages within five years because of a combination of rising temperatures, drought, population growth, urban sprawl, waste and excess.” Huffington Post 2007
- “Across the globe, water consumption has tripled in the last 50 years, and at least 36 U.S. states are anticipating some areas of water shortages by 2013, according to the Environmental Protection Agency.” Reuters 2012
- “At least 36 states are expected to face water shortages within the next five years, according to U.S. government estimates.” Sierra Club
Like any real zombie prediction, this prediction began as an honest living breathing prediction in 2003. After the drought of 2002, the General Accounting Office (GAO) was directed to figure out how the Federal government could help states meet their future water needs. Thus, GAO issued in 2003 GAO-03-514.
In this report they noted that no federal entity has comprehensively assessed the availability and use of freshwater to meet the nations needs in 25 years and that “forecasting water use is notoriously difficult.”
So, how did GAO calculate this prediction? They didn’t. The prediction was based only a compilation of web-based surveys of state water managers. Three important states didn’t even participate; California, Michigan, and New Mexico.
The report states that “even under normal conditions, water managers in 36 states anticipate shortages in localities, regions, or statewide in the next 10 years.” More important, 46 of the 47 water managers said their states are likely to experience shortages within the next 10 years under drought conditions.
I don’t have a problem with this report. There is nothing inaccurate or misleading (except someone slighted the south and mislabeled the map. Can you find the problem?). My problem is its continued use as a prediction. I believe it’s a flimsy prediction that is close to ten years old that appears more authoritative than it really is. Perhaps it is time for this report to die a nice death, buried in some musty government storage facility beside the Ark of the Covenant.
This prediction reminds me of some of the stories I am enjoying in Nate Silver’s book Signal and the Noise : why so many prediction fail- but some don’t. Basically he says we love to predict things, but sadly, we’re not very good at it.
A constant theme so far is that within this big data world we can all improve in our attempts at accurately forecasting the future. A fantastic book in the spirit of Freakonomics for anyone who wants to understand more about about statistics, about prediction, and even about themselves.
Wherever there is human judgment there is potential for bias.” -Nate Silver