I've been thinking about it, and I have a partial answer.
I think we can agree that a system where someone with good system mastery can make a character that is a little more powerful than a character made without focused intent (an "organically grown" one) BUT only a little - ie enough to make the careful character builder feel rewarded, but not enough that the game is dis-balanced. This would be a good place to be right?
Let's say that for warriors (rogues, monks, fighters, etc), a difference of 5 dpr (damage per round - and I'm just making up a number here, I'm not saying that it's the right number!) is about as much as you would want to see.
Let's say that each "best" choice increases your DPR by 0.5 dmg.
If you have a game with oh, 12 choices in your character design, you will have 6 more DPR than the "organic" character *IF* the organic characters makes only bad choices! (DPR wise). I think we're good.
But what if the game has 40-50 choices? Now suddenly, due to the cumulative effects of all these decisions, the optimized character is much, much better.
And that's why it's hard to do - if you have a rich, complex character design process, if choices are meaningful, and if there are a lot of them, the possible gap gets very wide.