>> Should those ideas just be forgotten, should they be translated/ simplified into something smaller and more manageable?
they should be added to your list of "possible solutions i'm aware of".
but do not pursue them until:
1. you need them for a game. then learn how to implement them.
or
2. you have nothing better to do than learn how to implement them (rather unlikely - there's always something else to learn in game development).
for actual ideas for a type of game, as tom sloper and sun and shadow are referring to, i jot down the game idea on a piece of notebook paper, and toss it in a manila folder with the other 50 or so game ideas i have on file.
but usually, if its a good idea, i'll think about it from time to time for 2-4 weeks, maybe make a few notes and do a little research, then stop my current project for up to a week and do a rapid prototype. it never even makes it into the game ideas file.
so much to build! so little time!