Melee Class Theory
It's been several years since 3.0 was released, and since then, there has been all sorts of discussion about the fighter. Arguably the most bland class, it has also been deemed to be relatively useless, given the lack of powerful feats. A wizard can easily outclass the fighter past 5th level, no questions asked.
This particular problem is resolved by the introduction of the Einhander class, which is essentially the material ripped from Book of Nine Swords and changed to taste (though those mechanics do seem somewhat overpowered). However, the scope of the problem is grander than that: the entire set of martial classes need retooling. Thus, I began the conceptual work on a new melee class design theory.
Rock - Paper - Scissors
Things are more balanced in threes, or so I would like to believe, and this counts not only for the Forces themselves, but also in the world of melee class design. The basic theory is rather simple (though it does get more complicated later); there are three parts to the whole, each of which excels against one and is weak to the other.
RPS is the perfect analogy. There are three types of melee classes: the rock, which focuses on defense; the paper, which focuses on mobility; and the scissors, which focuses on offense.
The best example of this is the Defender and the Einhander. If you pit an Einhander against a Defender, the defender will win by sheer attrition - the einhander isn't capable of regularly hitting the defender (due to high AC), and even if he does, the damage is reduced (due to high DR). The defender is designed to handle being attacked constantly, and since that's what the einhander does best, the defender wins.
This theory allows us to design melee classes with a "melee alignment," as it were, which opens up design space. It means that we don't have a generic fighter class for anyone who melees - it means that a melee combatant chooses a style of fighting and goes with it, accepting that it has strengths and weaknesses. It grants us versamilitude, and allows us to design different classes that fill those roles, giving players more options.
Dynamite and Arrow
At the stage above, however, the theory is incomplete. There are two other kinds of "melee" combatants, which are not covered by those descriptions adequately. The core Rogue, for instance, is not a melee type, but has intense damage-dealing capability. The archer is not a melee combatant, but his skills are still oriented around combat (but from afar, rather than toe-to-toe). Rather than shoehorn these classes into one of the other three categories, we instead give them their own categories, and expand the RPS analogy.
Rogues and their ilk - that is, lightly-armored classes that deal intense damage in the right circumstances - are the dynamite. Meanwhile, archers and similar long-ranged but still non-caster classes are the arrow.
To bring these two new groups into the theory, we group the rock, paper, and scissors into a group (referred to as RPS, here). Dynamite's ability against RPS is circumstantial: if dynamite gets the jump on RPS, dynamite wins. Without surprise, or if the fight goes too long, though, dynamite loses - it lacks staying power. Arrow's prowess is also circumstantial, but based upon range: if the arrow has range, it wins, but once the distance is closed, the arrow loses.
Dynamite and arrow suffer the same against each other, as well. Dynamite jumping on arrow, the arrow is doubly weak (due to not only getting jumped upon, but also from being in close range). Arrow against dynamite, dynamite lacks surprise, and arrow has range. Hence why the two are not grouped into the same category.
There is also the idea that a class can occupy two spaces at once. For instance, knights would be both a scissors and a rock. Again, this possibility adds complexity, but it also adds interesting design space.
The immediate question is - what is such a class weak to? There are a couple arguments that could be made. Knights, for instance, could be weak to rocks (because they are scissors) and to papers (because they are rocks), but these weaknesses would be lessened because they have two. Alternatively, the knight could be doubly weak to paper (because it is a scissors-rock). It could even be possible to have two classes for each hybrid role, each fulfilling the two possibilities.
This theory of melee class design allows a designer to pinpoint exactly what the goal of a melee-type class is, something that was previously a generally vague concept. Specifically, it clearly identifies several subroles that a melee class can fulfill, which is incredibly helpful in class design. Having a framework that allows one to specify a class or even just an ability as a "rock" gives designer, DM, and player a clear picture of how the class or ability is meant to function, and how best to put it to use.
That's the idea, anyway.