One of my professors at KU "created" RPM:
http://espn.go.com/nba/story/_/id/10...eal-plus-minus
With that in mind, I wouldn't place too much stock in it.
First, ESPN isn't publishing what goes into RPM. That doesn't necessarily make it a bad measure, but it raises eyebrows.
Probably the best article I've found about RPM is here, where they visit the APBR board to try to find the full RPM methodology:
http://www.hickory-high.com/is-espns...inus-for-real/
A couple of points from the article:
1. RPM uses data from the prior year to reduce noise.
2. RPM contains a height-based prior which boosts the defensive ratings of all taller players.
So in the case of Stiemsma, for example, his DRPM is boosted by his height, and influenced by his time in Minnesota. You can see where this can create problems.
Despite ESPN's tenacity, any "all-inclusive" metric to rank players (PER and now RPM; QBR in football) is going to be hugely flawed. It's just too complex a game to tie every factor into one number.