We called this PRL and tested this, doesn’t work well. PRL = Potential Retreivability Loss.
Also, we tried PSG, which is Potential Stability Gain. This didn’t work as expected either.
I strongly suspect the reason the testings were unintuitive is because we missed something, i.e. something was wrong with the simulations, but I can’t say for sure. Take a look at this thread if you’re interested: New Sort Oder: PSG
The linked discussion thread there has too many replies so you might want to sort by likes, and read the top ones.