3.75 Update General Policy

I've put up a few updates to 3.5 classes that weren't core classes and were inadequate in the 3.75 (or in some cases, the 3.5) environment. I feel that making my approach more clear may help answer some questions about the changes I've made.

Rule #1
Mastery Is Achieved When There Is Nothing Left To Take Away
If a mechanic never comes up or almost never comes up, I don't see much point to devoting the space to it. Likewise, if a mechanic is needlessly complex, I see no reason to implement it, especially if there's a simpler way of doing it.
Example:
Grace (Swashbuckler): Grace adds a total of +3 to the Swashbuckler's Reflex Save. This is a typeless bonus. It only applies when the Swashbuckler is unencumbered by weight or armor. Unlike having a good reflex save, it doesn't help the character qualify for Prestige Classes, and it only gives half the bonus, and then only during certain scenarios. This ability is narrow and convoluted.

Rule #2
Only Zombies Have Dead Levels
Using the Core classes from 3.5, Level 5 was all over the place. The Wizard and Cleric gained the ability to fly, to launch fireballs, and to create zombies. The Fighter got a point of base attack. The goal in 3.75 updates is that no class should ever have a "dead" level, i.e., a level with no class features. In general, gaining spells known doesn't count as a feature, but opening up a new spell level does. In the case of Classes like the Warlock and Wilder, where spells (or the equivalent) are learned infrequently, they do count as a class feature. Likewise, special spells that are granted infrequently (like the Exotic Learning on my 3.75 Wu Jen) also count. The difference is largely whether I feel what's given at a level is something to look forward to. Note that class features need not all be great; sometimes you get Smite Evil, and sometimes you get Trap Sense. This is largely done for balance purposes, and is balanced against the idea of "something to look forward to."
As a side effect, making sure every class gets something at every level also makes Racial Hit Dice more significant. I know a die of Monstrous Humanoid in 3.5 was really no worse than an odd level in Fighter.
Example:
Swashbuckler Dodge Bonus. +1 to AC isn't exactly earth-shattering, but it's nice to have. Plain and simple.

Rule #3
There Are Non-Casters Above Level Ten
3.5 was notorious for some classes simply not being viable above a certain level. The Epic Level progressions for Fighter and Ranger were somewhat naive, as even by level 15, a straight Fighter would find himself so far behind the casters that he could never catch up. While I don't want every class to be as powerful as a Wizard at level 15 (because few are as weak as a Wizard at level 1) I feel that every class should at the very least be able to do something neat, something that makes a player want to take it up that high. A factor here is that getting to level 20 is not something most people do (I like the idea that most people are levels 1 to 6), and reaching that level of personal strength should allow some pretty impressive stunts.
Example:
Dancing on Sabres (Extraordinary Style Crowning Achievement). This ability is not the most powerful ability, but it's really freaking cool. Nothing quite shows the effortless grace of a Swashbuckler like hopping up onto a foe's weapon and laughing at him as he tries to shake you off, chopping bits off him all the while.

Rule #4
Superman Is Not Level 6
By the same token, the first five levels are the most important to watch. These are the levels where casters should be the shakiest and fighters should still be watching their backs. Not only can a class that's too powerful at low level cause problems in a lower level game, it can cause problems when it gets multiclassed in a midlevel game.
Example:
My original draft of Samurai gave it an ability called Iajutsu Strike that gave him an attack dealing +3d6 on attack at will as a standard action. At level 6, this wasn't as bad; while the other warriors are making 2 attacks, he's still only making one, but the tactical abilities of a standard action showed me the reason this mechanic was difficult to balance was because it wasn't balanced. Hence I took the class in another direction.