Mechanisms that are First-In/First-Out (FIFO) (path independent) do not like variability and volatility (i.e., Jensen’s Inequality/Antifragility) as much as ones that are Last-In/First-Out (LIFO) (hence path dependent).
Take diabetes. We are discovering that diabetes is not (as we thought) the result of being overweight, rather the effect of absence of variation, not losing weight, not having periods of starvation (that among other things, clean up the fat deposit in the pancreas that is LIFO). So someone overweight who loses weight can be much better-off than the same person a bit thinner at stable weight.
—
There is plenty of research in diabetes hinting at this from many sides but nobody tried to put a systematic mathematical apparatus on it. Though the math is not trivial (because of path dependence), I was able to play with it with Monte Carlo analyses.
—
Note that the Russians have known that for over a century.

http://www.ncl.ac.uk/magres/research/diabetes/reversal.htm

See also (easier read)
http://www.dailymail.co.uk/…/I-reversed-diabetes-just-11-da…

Reversing Type 2 Diabetes – Newcastle Magnetic Resonance Centre – Newcastle University
Our work has shown that type 2 diabetes is not inevitably progressive and life-long. We have demonstrated that in people who have had type 2 diabetes for 4 years or less, major weight loss returns insulin secretion to normal.

via Facebook