04 August 2014

Live Long and Prosper

The stats I was brought up on stressed that one should never, ever predict/infer beyond the range of the data. Nowadays, doing so has become far more than a cottage industry. As stressed in these pages, the only circumstance where prediction past the end of the data is (nearly) reasonable is for physical processes, which have reached stability. That is, the exogenous forces have become static.

Human processes, especially science and engineering, violate such requirements in spades. So, it was with some amusement that this post came via R-bloggers. Life expectancy changes, increase or decrease, are due to humans' ability to fool Mother Nature. It wasn't just the passage of time that life expectancy (at birth) increased by a couple of decades from the mid 1930's to today. These additional years of blessed human-hood came about as the result of improved public health, medicine, surgery, pharma and the like. Not to forget breaking the tobacco habit.

Assuming that there exist, by definition, more such interventions to move the whole of a nation's population to greater life expectancy is just stupid. No other way to say it. As it is today, new interventions are increasingly costly, and will, if not immediately then in short order, be restricted to the 1%. The needle of national life expectancy won't move even a wiggle. We've conquered the causes of widespread less-than-adult mortality. And we're not making much progress with geezer mortality. We've reached the asymptote. Too bad the quants haven't the sense to look out the window to see whether there's rain or shine.

No comments: