I'm going to make another suggestion here; either approach may be valid for you.

Your pattern seems to be that you begin something, hit a roadblock, can't get beyond it, then give up and try something else. Rinse and repeat.

Unfortunately those roadblocks seem to happen quite early and with quite findamental things, but yet you obviously want to do this kind of programming.

So my suggestion is that - instead of starting from scratch - why not take an existing open source game engine - one of the Quakes perhaps - and bash at it for a while?

You won't be dealing with the best, cleanest or most nicely written code in the world, but you will be dealing with code that works. With code that has already solved the problems you seem to get stuck on. There's no need to get too ambitious at an early stage, just trace through it, learn how things are structured, get a feel for what's done where, and remember that it's often going to be nasty gnarly code, so maybe try to rewrite some of it to be nicer, with the safety net of always being able to roll back if you mess up. Build up some experience that way and it might help you with some of the things you get stuck on in your own work.