I've had a lot of success at Hi-Score with running game design workshops. The basic formula is always the same: I create some initial constraints, and then let the participants loose. The constraints make the creative process easier, because the designers are forced to start off asking themselves questions, and asking questions about your game is a great way to narrow down what it is that you're actually going to build.
One of the easiest kinds of constraints to make is a thematic constraint. Your game can be anything, any style, but it must be about frogs. Or rain. Or paint. Or whatever. CMU ETC's experimental gameplay project did this.
In my game development class, the project I just handed out today asks the students to make a game on the theme of Light and Dark. Including two opposing themes joined by the word "and" does a subtle, interesting thing: it forces designers to ask, how do the two themes relate within the game? Are they opponents (Light versus Dark, e.g. Archon)? Are they collaborative (Player wields the forces of Light and Dark together)? Are they orthogonal (There is a Light World and a Dark World and the player must navigate between them, e.g. Zelda: Link to the Past)? You will get radically different games based on how the relationship is interpreted.
I expect you could have similarly interesting results with any two opposing forces. Design a game about Nature and Technology. Or Land and Water. Or Cheese and Chocolate. Anything to force the designer to start asking a question about the theme itself.
By sheer coincidence, my colleague Darius recently blogged about another way to present a constraint for a game design exercise: take a controversial, famous quotation and make a game to demonstrate the idea (or disprove it, I suppose). Could be a great design exercise for a serious game developer. I think this is also a great method because it forces the designer to ask two questions: (1) do I agree or disagree with this quote, and (2) how can I express my opinion through gameplay?