These days, whenever I think of a game idea, it seems to revolve around a toy. Some system that lets you do a wide variety of irrelevant but interesting things.
For example, I just built a silly little prototype for a game based around building weird steampunky machines on the outside of a weird steampunky space station. Lots of levers and gears and winches and stuff.
The algorithm is kind of fun, you can build a wide variety of things with relatively few pieces.
But it's not a game, and that's an important thing to remember. When people say that Sim City is a toy, not a game, I have to disagree. The game is to build a functioning city. While you are not limited in your approaches, the simulation offers very clear guides. You get smacked down by traffic, pollution, power, sanitation, fire... all of these factors that you need to manage, not to mention the economics.
It's an open-ended game, sure. But the algorithm is very strict, very limited. The goal is very clear. If you call it a "toy", you're saying that every open-ended game is just a toy... I guess if that's what you want to say, that's okay, but I disagree.
See, a toy would be a box full of building-shaped blocks that you could plunk down to your heart's content. No zoning, no economy, no rules. Just doing whatever you want.
This is an important distinction to me, because recently I've been making a lot of toys. I've come up with a lot of rather nifty little rule sets that let you do a lot of interesting things... but they have no external factors. There is no goal, open-ended or not. There are no limits, and therefore no inherent goals to overcome.
It's important to have external factors when you're designing a game. Something to measure success and failure with.
Right now, I can't really figure out a way to make the steampunk machine thing interesting. There's simply nothing to do. No matter how miraculous your machine is, it accomplishes nothing. There are no goals.
I can't really think of any goals for this particular game, none that are compatible with the engine. So I think it's important to think up goals before you really come up with an engine. The goals can be open, like Sim City, or they can be closed, like 99% of other games out there. But the gameplay has to interact with external factors, or it's just plain boring.