Polynomial extrapolation is generally considered to be bad, isn't it? And is there any intrinsic reason to believe that population growth would be polynomial rather than exponential or some sort of modification thereof?
Agree, but I was pleasantly surprised by how reasonable the estimates seem. A straight exponential would seem to be worse. Maybe you can suggest a more sophisticated function, and I could do a curve fit on it?
The 50-years-hence estimates come with a massive margin of error and I suspect that the 2070 figures will end up to be very different from the numbers calculated here. But I might be wrong, and there is one good way to check this out (apart from waiting 50 years). I'd suggest to Dan that he takes the population data from 1920, 1930, 1940, 1950, 1960 and 1969, then uses his polynomial tool to "predict" the population fifty years later, in 2019. If that exercise produces a result close to the real present-day population, then he might be onto something!
As a wise man once said, "It's tough to make predictions, especially about the future". No doubt these predictions will be wildly inaccurate! Maybe the next great entrepreneur will be born in Oregon or New Hampshire and totally reshape migration patterns. Maybe the U.S. will cut off immigration (or increase it). Maybe someone will discover a cure for aging. Who knows?
Good suggestion. Anyway, real world stuff determines migratory patterns. (Jim Crow, the rise and decline of the Fordist mode of production, and the increasing availability of home air conditioning would be three pretty disparate 20th Century U.S. examples.) That climate change may influence migration in the next fifty years is a pretty good guess, but generally, hindsight works better.
Estimates as recently as the turn of the century have turned out to be way off in many states...they had Washington DC down in the 400K range by the early 2020s, and instead its nearly 700K.
They also didn't anticipate the North Dakota oil boom...
They also didn't anticipate the North Dakota oil boom...
I tried using linear quadratic and cubic estimation. Non of them is much better than the others, however the cubic the worst.
Here the result of the quadratic (and the reality):
1 California 57.0 ( 1 39.8 )
2 New York 26.1 ( 4 19.9 )
3 Texas 22.9 ( 2 28.7 )
4 Florida 22.5 ( 3 21.3 )
5 Ohio 19.0 ( 7 11.7 )
6 Illinois 17.4 ( 6 12.8 )
7 Michigan 16.1 ( 10 10.0 )
8 New Jersey 14.0 ( 11 9.0 )
9 Pennsylvania 13.3 ( 5 12.8 )
10 Virginia 10.4 ( 12 8.5 )
13 Georgia 8.9 ( 8 10.5 )
15 North Carolina 7.7 ( 9 10.4 )
I think it is quite bad. 8 among the first ten was already there in 1970, however there were no chance to predict the other two.
Or that Florida might be too expensive to live in, etc...