I've just got finished getting my initial version of "D&D Thoughts" down on paper (or rather down online), and I wanted to switch gears and get down what I think the Next RPG should be, which just so happens to be the system I've been working on, but I want to go a little more into what exactly are the problems that I've attempted to fix.
Most RPG both tabletop and computer/console revolve around the players killing essentially legions of enemies, or even if they lean toward fewer battles the also tend to lessen the impact of any given round, by making the damage sustained in a given round a small fraction of a creatures overall health.
On the opposite end of this spectrum are the war-games where they are dealing with possibly 100 vs 100 opponents, where they need much faster resolution, but these games are not concerned with the contributions of individual combatants, they are all about squad vs squad rolls...there are simply too many units on the field to place too much focus on individuals, but they do demonstrate how deadly combat can be.
The systems that seem to have the correct focus are the tactical skirmish systems. In these their is enough troops on the field that they can't be over concerned about allowing individual heroics to play too great a roll, they minimize the offensive and defensive rolling to get to the resolution as fast as possible, they seem to have the correct amount to deadliness in them to get to the final resolution of the battle as quickly as possible, and yet the stronger side does typically prevail against the weaker opponent.
This is what I think should be pulled into more RP system mechanics. Too often the attack rolls miss or are blocked completely round after round, with in my mind seem like unnecessary dice rolling, the purpose of combat is to come to a resolution and move on. When the hit points or resistances of the defenders are such that it's going to take 20 rounds of rolling to determine who will win, than something is wrong.
I believe a good RP system should adopt a skirmish battle mechanic, what this does is combine the best of both worlds as far as individual contributions and getting to overall resolution. Combat needs to be deadly for it to have any meaning, so there's nothing wrong with a weaker opponent beating a stronger one...there's a lot of things that can be considered in a battle. The dice attempt to make a guess at the probability, but once the dice are determined then the outcome needs to be determined by their result.
Heroes will have special abilities that allow them to cut through some of the randomness of battle, the odds have to be in a hero's favor or they would never engage, after all how many times can one realistically beat a 50/50 scenario.
This takes me to my second point, which is "Damage hurts". Knowing that combat is deadly, but heroes have their ways of bending the odds in their favor, however lasting damage needs to be another equalizer in the grand scheme of things. When five minutes after a battle everyone is back up to full strength it takes away another whole aspect of the game.
Also if damage comes back too quickly, it completely negates the point of weaker opponents to be encountered. If a creature has almost zero chance of killing the heroes, but their damage lingers, now the opponent has a purpose...the party may have to rotate around who is going to be defending if they want everyone to make it to the end. It becomes more of an endurance contest, without lingering damage it completely removes these types of adventure scenarios from play.
Lastly, rolls must be made with a pool of dice, its the only way to account for both extremely difficult and extremely easy tasks. It's tough to account for odds in less than a 5% chance accurately without using a pool of dice, even d% most often are really only a d10 (if you have a 45% chance of success and if the ten's place is anything except a 4 you don't need the second dice, or if the 00 = 100 then you always need to roll the second die if the first comes up a 0).
When a task is easy the more dice that are rolled puts the odds of rolling average at a higher percentage, for example typical attributes are randomly generated using 3d6, which means only 1 in 216 will have and 18 or a 3, but your chance of having at least a 10 are about 50/50.
The addition of exploding dice (when a dice rolls the maximum number it gets to be rolled again, with continual rolling allowed so long as it rolls maximum) allows for those extremely difficult rolls to be made even when the chances of success are very small.
Dice pools also aid in contests where the warrior may be rolling 6 dice vs a wizard that may only be rolling 1 or 2 dice, but even with that there is a chance (especially with exploding dice) that the wizard might be able to beat the warrior in a test of strength. In which case you could likely chalk up the wizards victory to a well timed distraction that gave him the victory.
These are all things that all versions of D&D and Pathfinder fail at, which is ultimately the reason why I've felt the need to craft my own game mechanics to take these and more into account. Ultimately, I felt the need for dice pool based skirmish mechanics that allowed for fast and deadly combat resolution with possible lingering effects from encounter to encounter.