If you haven't been providing support here for very long, keep in mind that what looks great from one point of view may have serious problems from another point of view.
Having very "narrow" forums does not necessarily solve the problems you describe, at least not perfectly, which means you have to look at the disadvantages too. It's a balancing act.
Newcomers still pick the wrong forum from time to time, and the more forums there are, the greater the chance they'll do so.
If you make the forums too narrow, then some more general questions get asked in the wrong place too easily. Someone may not have a preference for how they solve a problem, they just know they're using the NDK and have available all the tools it provides. They may say "How can I use Cascades to do X?", and get an answer along the lines of "You can't, but you can just resort to pure native code by doing Y.". Then it evolves into a whole discussion involving "pure native" code, and nobody knows where the discussion should belong.
Then there's the issue of critical mass. Many people tend to pick only a few forums, or one forum, to follow. If you divert all the OpenGL stuff, and all the Cascades stuff, then the main NDK forum may have relatively few regulars, so the quality of the support drops and people complain (as I often see them do in other forums) that nobody answers their questions in the BB support forums. We can be spread too thin...
I'm not saying there shouldn't be one or two more forums, however... but I am saying that it would be wise to move slowly, after we've had ample time to monitor things and identify all the problems and needs. I'm also saying that those of you who don't do as much support in these forums as others do might want to consider the difficulties involved in (for example) subscribing to too many different forums, constantly having to ask moderators to move threads to a better forum, and making it harder for people to find the more general information when they don't have a narrow question. It might seem to make it easier on you to have one "clean" forum that suits exactly your needs, but you won't like it if, as a result, there is almost no one else there to talk to.
Also, perhaps the split shouldn't be so much along technical lines. Maybe a "game development" forum would be helpful.
It might be a good idea to see what focus different people have before deciding how to slice things up. I often care far less about what platform I'm using for a particular problem than just how I can find *any* solution that might work. For example, I happen to want to do some nice anti-aliases vector graphics stuff in a Cascades app. The solutions may involve something in Qt, or future stuff in Cascades, or maybe OpenGL, or a third-party pure C library that can be ported to the NDK. I'd prefer not to have to "shop around" through three different forums just to try to get an answer to that.
