Pretty much. Papa Mole says that you can play the IG as a fighter or a diplomat (which to me is the very essence of role-playing) but he refuses to accept it because the fighter path in a military faction gets better reward and treatment.
I think you're confusing an important issue here. It's totally okay for the
character to get lesser rewards and treatment in the story, but it's not ok for the
player to be shortchanged playing the game. And who plays the character building and optimizing game, if not the player? If you want a player to roleplay a character doing suboptimal things in the story, at least reward him as much in terms of player resources (SP). Otherwise you're just training your player to play archetypical characters / characters for which you've decided the path was optimal. Conditioning, man. No carrot no fun.
[And... this made me jump in and try to troubleshoot AoD from a conceptual design perpective.]
There's another thing that follows from the player/character distinction. I understand the goal is to make the game as realsitic / believable from the
character point of view, but that shouldn't come at the expense of the player experience. Requiring the character to have some level of competency to be able to do something is a great story/"in-character" goal, but you've to to think about the player here, because he's the one playing the game. A player needs to know what he gets for putting skill points in a skill, which is a system gaming / "out-of-character" decision, he needs to know what's the use. And one of the problems with skill checks and requirements is that the player never knows what requirements are coming, so he makes character choices blind, or instead he does a lot of trial and error, failing a lot and reloading so he can have that kind of precognition that allows him to plan his character build so it's viable for the character roleplay he wants to have. I don't know about you, but both options seem bad from the player experience point of view. Either I make choices blind, and find out eventually when I fail that my character build doesn't support the roleplay I had envisioned, or I reload a lot to get the character I want, making the game something tedious. That's really punitive. Sure, some will tell you it's all fine, but they're the players that will replay the game many times and want system mastery. But IMO system mastery should never come at the expense of player experience.
There's another way to show the problem. You have 2 design goals that in the way they're implemented work against each other: Skill competence and C&C. On one hand, you want that some options be only available for the competent, so you put in skill requirements. The consequence is that you've designed multiple paths in the game, but for a player playing it through, only a small number are available at a time, directly related to their character build. On the other hand, you want C&C, you want the player to try things, and for that you need a breadth of choice, and you need to facilitate experimentation with the character role. Yet the limited skill point availability make it so that you can't really experiment much, play out of type and slowly redefine the character, because the system is not forgiving.
So the design is working against itself. If roleplay and C&C are main goals, then the character build system must be subordinate to them, working as a facilitator to allow the player to make choices that define the character. The character build should be flexible, and react to the roleplay. But you've got it the other way around: the system dominates RP and C&C possibilites, by imposing limitations, with skill choices and skill requirements, that reduce the number of options available.
So, as a designer, what do you do, so that each part of the design complements the other, therefore making the player experience coherent instead of frustrating? You have options. An easy way to do some of it is the Blank Slate, and doesn't require changing your system. Start the character with 0 skills, and lower skill requirements. This way all paths are open to a player at the beginning, and he gets to choose how he wants to develop his character during the game after he's seen what it's like and can anticipate a little. Or make the development dependant on actions taken. Then the choices you make literally define your character.
But maybe that's unacceptable for the in-world believability and you want some level of competency at the start. That's harder. You need to keep options open for the player, so he feels good as a player about the choices he makes (the character can be miserable, that's cool), and at the same time make competent characters better at what they do. With a skill level requirement system, non-competence is punitive, options-limiting. To keep to your design goals, you either gotta limit the options-limiting impact of competence, or go for another way to make the character's competence relevant in the game, that is, instead of punitive, rewarding.
Let's start with the first way, less option limits. I'd chop the number of skill levels, and go for something simple: a layman-apprentice-master system. 3 levels of competence, so you can implement a consequence for each, making all levels relevant in the story and all character build choices useful for the player. A player will know that each point invested is significant, and 3 levels makes for a lot less guessing about future requirements. This also allows you as a designer to create non-total-fail paths for layman characters: the skills levels are in low number enough that you can design a satisfying result for each.
But there's other ways, aside from skill level requirements. Drop those, and instead, give something to the player for having invested in the skill, that way investing in a skill is a reward, yet non-options-limiting. How? The new Torment has Effort, so your skill is a resource you can spend to get more out of an interaction, but that's still tied to skill requirements, we'd need to uncouple it. Maybe something like this, for example: instead of having a Bluff skill to bluff more perceptive NPCs, instead make bluffing, intimidating, etc available to everyone, but add an Empathy skill that gives you more information in dialogues to allow you to better make your in-character choices/roleplay (spend Empathy for info, or just info depending on skill level), so your character has maybe a hint that bluffing won't work in this circumstance with this NPC. Maybe you can also combine with requirements, and a master in Empathy opens a way to bluff him. End result: options are still available to all, but the player gains something for his spent skill point, and the character is visibly more competent. Same for combat: instead of having a better combat skill making you have a better chance to hit (actually, being viable as a combatant), make the skill give you new moves and abilities. Hard and tactical combat doesn't require a skill level requirement system. It's boring for the player to not be able to hit, but it's not boring to not have a purchasable ability. There's the issue that more combat abilities makes combat easier, but if playing a combat character, you'll also get into harder fights. And sometimes the fight's easy and you get to show off your skill. It works out.
I could go on... but this is the line of reasoning I'd follow. It would also solve the problem of hybrid character builds, since instead of tailoring the game to certain character builds and gating content with skill level requirements that are either too easy for a specialist and too hard for a hybrid character, every character build point / resource can be used in a positive way, instead of having the design assume that not having the requirement limits your options.