To respond to the original post, the sort of things I do to come up with ideas are:
- Recreate / take inspiration from the feeling a game, or a portion of the game, has given me.
- Experiment with changing aspects of an established game / genre.
- Recreate / take inspiration from an aspect of a game.
- Recreate / take inspiration from a single level of a game.
- Experiment with control / input methods.
- Play about with physics (more fun if you can do your own).
- Do something I've never done before (this might sound a bit obvious).
- Elaborate on something I've done before.
- Just sit and code stupid, simple stuff (the coding equivalent of doodling).
Pretty much everything I've posted on the forums (and a lot of stuff I haven't) has come from this kind of thinking.
It really depends on what you're trying to learn / achieve and what interests you. I like to make arcade style games and like playing about with physics so most of my ideas are about getting things to move about on the screen and player controls.
Just to touch on the "Acrogorpon" project that Chris linked to: it's obvious that DL187 put a lot of effort into the game but I didn't find the comments on Steam Greenlight surprising. I thought the art style was generic but then my own FPS game is purple, so maybe I shouldn't be commenting. It must've been quite a harsh reality check (a bit like Fallout's Kickstarter project) but I admire the attempt. I hope that he/she has taken the comments and sought a different target market or used it to galvanise a determination to try again, to either improve or try something different.
As for fun, making games isn't fun in the normal sense of "let's go out for beers" fun. It's a kind of self inflicted torture and the reward is the small triumphs of figuring things out and the feeling that comes with creating something that otherwise would not exist.
one of these days I'll come up with a better signature
