I just read about this, and I don't know when I'll have time to try it myself. Actually, I don't even know if it makes any sense, but it aroused my curiosity.
Have you tried it: Reverse audio, and compress it more heavily than you'd ever compress it when unreversed. Then reverse it again, back to the original position.
The assumed benefit of this is that the heavy compression doesn't destroy transients when audio is reversed. But of course, the result must be a little weird... My first thought was that of course the transients are destroyed, just in a diffrent way, when they're left in the release tail.