In October 2010, I published a <a href="http://alum.mit.edu/www/tchow/monthlyprob11523.png">Monthly problem</a> that introduced the concept of a <i>fair permutation</i>, which is a permutation $\pi$ such that for every $i$, either $\pi(i) > i$ and $\pi^{-1}(i) > i$, or $\pi(i) \le i$ and $\pi^{-1}(i) \le i$.  Equivalently, every cycle of $\pi$ is either a fixed point or an alternating cycle of even length, meaning that if $z$ is the largest element of the cycle, then
$$z > \pi(z) < \pi(\pi(z)) > \pi(\pi(\pi(z))) < \cdots < z.$$

I don't think anyone has studied fair permutations explicitly before.
However, 
a permutation $\sigma$ on $\lbrace1, 2, \ldots, 2m\rbrace$ is said to be a <i>Salié permutation</i> if for some $r\le m$, $$\sigma(1) < \sigma(2) > \sigma(3) < \cdots < \sigma(2r)$$ and
$$\sigma(2r) < \sigma(2r+1) < \sigma(2r+2) < \cdots < \sigma(2m).$$
These <i>have</i> been studied before, and one can show by generatingfunctionology that the number of fair permutations is twice the number of Salié permutations.  This is very suggestive and hints at a close connection.

><b>Question</b>: Can one construct an explicit 2-to-1 map from fair permutations to Salié permutations?