Slaad, your comments about at "shifting bell curve" wrt dice pools intrigues me. You have a link or anything to a discussion thereof?
Well, I rail on this point over at RPGnet from time to time, but considering their suck-arse search feature, I doubt I could pull it up.
Short take: I think that being able to understand the odds of the dice roll are a substantial boon to the GM of any game where dice roll contribute substantially to the course of the game. That is, most gamist or sim games (in GDS sense) that run with little fudging.
For games with one dice, it's pretty trivial to understand the odds.
Games using multiple, but fixed, dice are more complicatad "triangle" or "bell" curve, but you can usually come to a pretty good understanding of the odds in short order.
But when you have dice pools, this situation becomes hopeless because every dice you add represents a DIFFERENT bell curve. So to understand the odds, you have to get a knowledge or feel for multiple bell curves, one for each dice combination. That makes it difficult for the GM to develop a feel for the odds, and multiple bell curves are difficult to draft and memorize as a play aid as well. (I used to chuckle at oWoD GMs who'd tell me how simple there game was, but carried around cards with them with the odds.)
As far as dice pools go, simple binomial* with difficulty scaled to successes (instead of changing the target number, as in nWoD or Burning Empires**) can be TOLERABLE in this regard, but other variants, like additive, "binomial with shifting target numbers", or "roll and keep" variants are very difficult to determine or learn the odds for.
* - Binomial is the sort of system where you count "successes", like WoD.
* - But note that luke puts a (rather simplified) table right in his book.