I think that one of the main problems the rogue has is his mana gain ability. It doesn't incentivize the right things. The rogue gains mana for kills, right? But how does he get those kills? Using his skills of course. But how can he get the mana for those skills? From kills. The problem with this is that if he faces mobs that are too hard to kill e.g. bosses, he's stuck. He has to resort to his mana generating skills which take up precious time. It also makes those particular skills indispensable.
The fact that the rogue loses mana over time means that he doesn't have an incentive to lay low and wait for a kill. Just like the warrior, he has to attack to get his mana. Unlike the warrior, he has to kill to get it. This gives a strong incentive to invest in strength so he can kill quickly, which is not what we want to emphasize. We push the rogue to do things he doesn't want to do.
At first I thought we should just take away the degenerating mana and let the rogue keep his mana that he worked so hard for. But then there are some skills in the rogue's tree that are specifically built to freeze mana reduction, and those would have no use.
After thinking about several ways to fix this issue, I've come up with something that at least on the surface, sounds interesting. What if the rogue gains mana with time like a magician? Not nearly as much mana of course, but some mana relative to his fixed mana of 100. This reflects his preparation time -- the more he lays low, the more he recharges.
The twist is, that when the rogue gets hit, he loses mana per hit. This emphasizes several things: 1. The rogue should avoid being hit ie. high dexterity is a good investment. 2. The rogue should stay out of combat as much as he can, specifically against multiple foes. Fighting one foe will only reduce his mana a little. Fighting many foes will reduce his mana a lot. 3. The rogue is best fighting with a partner to attract aggression, unlike the fighter who becomes enraged (ie. receives mana from being hit) the most when he's alone. Finally, since this becomes the rogue's way of losing mana, his freeze mana loss skills are still useful: they simply prevent being hit from causing mana loss for a certain amount of time.
I like the idea of adding a shuriken to the rogue, but I think it should be an expensive option mana-wise to prevent him from becoming a purely long-range class, and dominating all of his other skills. The shuriken is something you'd shoot once or twice while approaching an enemy, right before you engage in hand-to-hand combat. I think the variation on mana gain here will make his combat different enough.
Assuming this is a good idea, this also gives me ideas about what I could do for DC. One can split up the classes into those that like fighting and being hit, and those that don't want to fight too much and want to avoid being hit. For the fighters, mana regen will decrease mana or at least be neutral (depending on the specific class). For the magicians/rogues, mana regen will increase mana and being hit will decrease their mana. Like the rogue, magicians also don't like being hit -- this reflects the difficulty in casting spells while being attacked. Mixed classes (fighter/rogue or fighter/mage) will neutralize each others' advantages and disadvantages. This also solves the problem of mana regen effects stacking in a pure class like Mage. Sure, they get more mana regen, but they also lose a lot more mana when they get hit.