Huh... building large worlds is long run job, and if you want to finish you've got to be really skilled.
There are several very important things to do in making open-world (\~large scale) game.
You gonna need tons of meshes, textures, sounds, etc. Preferably from single graphics team, because everything must "fit" into world. This brings you to point 2.
The problem is to design whole stuff. F.e. We're working on RPG game, for 1x1 km of world (and the further the development goes, the more convinced we're that world won't be bigger than very few squared kilometers) we have dozens of papers, sketches, etc. E.g. to even know how that world should look like.
The biggest problem. We have advantage of all being (except me, I can't judge myself) skilled programmers one of us has even more than 30 years of experience. And we have released (and we're still supporting & releasing) successful working applications. So we decided to write our own. Handling huge world means that almost everything in engine is dynamic = needs to be optimized. We have been working on technology for few years now, and it is capable to work with large worlds today. After some 3 years (or even more, I really dunno now). Some stuff is still work in progress though (AI, physics) - because we're trying to add new stuff.
Summed, you either need few skilled people, or at least \\$1M (note it's OR - so these two conditions aren't excluding each other), or maybe even more.
Start with small 2D things like Pong, Mario, then try Tetris, etc. It's a great practice (I even thought of releasing few of my C# (C++, etc.) practice-programming applications that are actually small games or even useful programs (Including alarm clock on notebook that is sometimes needed!).
Just don't stick to one project in the beginning, you'll run out of it - do rather smaller quick things, until you're enough skilled. It'll make you ninja. :ph34r: Then you can search for buddies and start on something big.