Gee, what a difficult question... Sometimes, what I do is I get out my notebook and begin planning things out. The first thing I do is make a flowchart to get everything laid out. Or, if I'm really confused, I'll sit down and make a list of all of the things I want in the game. (I've found that planning out what you expect in the game is of EXTREME IMPORTANCE -- otherwise you'll get "feature creep" and continuously add things and and things until you realize that you aren't getting very far.)
Then, after the flowchart, list, and other basic planning, I'll begin sketching all sorts of items -- characters, weapons, landscapes, and collectibles. Then I attempt a sort of sketch of a level, which really sets up mood, lighting, and general look of the game. I can then branch off from there to other levels.
After that, I switch hemispheres in my brain and begin working things out logically, because when I look at my plans, I generally think to myself -- HOW ON EARTH AM I GOING TO DO THIS? At this point I'll make a testbed so I can figure out how characters move and such. Then I go into more detail with levels and involved modeling.
However, what's odd is that one time, I completely ignored this entire system, and created one of my best games yet just by sitting down and designing. Hmm... I'm still up in the air about which is the best. Personally, I would give much more planning to large games with multiple levels, but arena battles barely take any planning at all.
Then, after the flowchart, list, and other basic planning, I'll begin sketching all sorts of items -- characters, weapons, landscapes, and collectibles. Then I attempt a sort of sketch of a level, which really sets up mood, lighting, and general look of the game. I can then branch off from there to other levels.
After that, I switch hemispheres in my brain and begin working things out logically, because when I look at my plans, I generally think to myself -- HOW ON EARTH AM I GOING TO DO THIS? At this point I'll make a testbed so I can figure out how characters move and such. Then I go into more detail with levels and involved modeling.
However, what's odd is that one time, I completely ignored this entire system, and created one of my best games yet just by sitting down and designing. Hmm... I'm still up in the air about which is the best. Personally, I would give much more planning to large games with multiple levels, but arena battles barely take any planning at all.