I've been finding that many of my projects fall into the abyss of my "game mechanic tests". That is, they aren't really "games", but rather sandboxes where I try out various things and find some mechanics that would be fun.
Now, I'd like to actually create a game out of some these ideas. Since so many of my "games" end up turning into additional sandboxes when I get distracted by a new idea("whoa, that's cool", and suddenly the whole game completely changes.. twenty times.. sometimes back and forth between several competing ideas), I'd like to actually plan out the mechanics, how they will fit together, and altogether how the game will work as a whole, even though I know I still need to be semi-flexible for when things play-test differently than expected.
This is compared to my usual strategy of, "Let's just get started and then I will build a game out of whatever I create" mindset.
What techniques are used for planning out a game idea (even if just roughly), before actually just jumping in? Is there specific software which helps? How do I go about managing the planning phase in general and keeping things up-to-date if things start to change?
See this great answer on a separate question for the type of actual design I would like to work on for my game, opposed to the programming focus I generally approach things with.