I favour more complexity in some areas, but overall more realism or elegance of mechanics. I do not particularly care for a large number of classes -- the core classes have always been enough for me in D&D, for instance, and I typically don't even use half of Rolemaster's core classes -- and in the case of
Dragon Age, I don't see a good argument for having more than three.
Considering that there is already a system for injuries, that is what I would most like to see expanded on -- criticals dealing injuries, and laying people out when there are enough of them. I'd love to see characters even being outright killed, but I'm guessing that's not going to happen. In fact, greater lethality of combat combined with more ways to build an effective warrior is what I most want -- particularly being able to build an effective warrior
without relying on activated abilities, which to me, muddles the line between playing a warrior and playing a mage with warrior-spells.
As an example of what I mean by more elegance and realism, less complexity: weapons choices. I do not think weapon choices should be restricted between the classes, nor do I believe that general combat styles should be, or armour. I believe that any class should be able to use any item, although perhaps not well (only mages could use mage items for magic, for instance). I'd rather see as class-specific abilities more of how once is trained in the weapon, rather than if one can be effective with the weapon at all.
DA:O did much better at that than DA II did. Simply returning to DA:O's combat would be a great improvement, but there are still more improvements that could be made.
It is possible the developers could be underestimating the casuals, but then one could look at the sales of Skyrim (16.51 million and counting) and make a case that they are not. Skyrim streamlined the system that was used in previous Elder Scroll games. In fact Skyrim doubled the sales of Oblivion.
Did Skyrim do better because it streamlined the complexity but kept the depth?
That would imply that a very large number of people bought the game solely because they removed classes and stats, and that is not something I am inclined to believe easily without proof. I think it more likely that the
Elder Scrolls is a well-enough established series that, particularly with good advertising, any new installment is likely to do quite well unless they make a massive misstep, and that didn't tick off enough people to count -- and
Skyrim was the first game in the series that worked equally well on all platforms (at least, from what I have heard the previous ones had some issues on consoles; I've no first-hand experience of anything but their performance on the PC). I'd be more inclined to put it to that, combined with the fact that games in general have sold more copies over the years.
Also,
Oblivion had graphical requirements that were enough in advance of the general PC of the time that many people probably held off on buying it because of that, whereas
Skyrim didn't. There are all sorts of factors that go into this. Lack of classes may have been one for some people, but I would be surprised if anybody bought the game purely because it did not have them.