Starting a New Design
I have a document, where I write down any ideas I have for new games. When I want to start a new project, I just have a look at this document, and pick something.
Inspiration
I'm constantly inspired by things I see in other games. I think "This is great, but I could do this better." Sometimes I just want to take one piece of a game, and make a whole game about it.
I've also used a process I call "scaffolding". I just start with a prototype that's quite close to another game. Sometimes, I'll map out the structures of the original game. I could, in theory, just start with the entire original game. I work on it, and after a few dozen revisions, it doesn't resemble the original game at all. Have I copied the original game? Well, no, because it's not there any more (except maybe a few generic core pieces).
The "initial thoughts" document
When I begin work, I open up a new blank document. I just start writing my thoughts and musings, as though it's a letter to someone. (It's also instructive for myself, later).
My adventure game document begins:
I want to create a game where you go adventuring, and travel across a land. I want to evoke feelings of exploration, discovery of items, survival, fear, and empowerment.
I talk about various big picture things for a few paragraphs. After a few lines, I start digging into how the game should function.
Get around movement lack of access by giving the player "go home" and other potential long-distance travel cards, plus lots of short ones. Maybe just a hand of 4x1space, 2x5, 1xanywhere. Can improve the hand too. Hand is presumably recoverable, or the player gets new cards somehow, and they're single-use.
By the end of the document, I'm laying down data and the turn structure, and it's time to go off and actually create the game pieces.
Ask the fundamental questions
This is the point at which you make the fundamental decisions about what the game is. This can't really be changed, without turning your game into something else.
What does each player have? Resources? A hand of cards? A player board?
What is the interaction?
What are the actions?
What is the turn structure?
What is the objective?
Do the first revision
Congratulations! Your game is half way to complete. The other half of the work will take 100 times longer.
The form of the game is cast. However, it's still possible to make significant changes at this point, if you keep the same structure.
If you find yourself wanting to change the structure of the game, you don't have a game. A game is its structure.
All my good games have been fun (even if broken) on their first play. If it isn't fun, abandon it. It doesn't matter how good an idea it is.
Approaching the middle
Early revisions are steps forward in the fog. I don't know where I'm going. All I want is to be able to take the next step. From there, I'll hopefully be able to take the following step.
This is a disconcerting phase of development. The game may still die here, which means any work I do has a decent chance of being thrown out. I find it hard to motivate myself to work on something that might amount to nothing.
Eventually, hopefully, the game settles down. It's good enough that I know it will survive. However, the game is not what I imagined it would be. I just took steps in the fog, and I think I'm going somewhere good. Always let your game be what it wants to be. Don't try to force your pre-existing ideas onto the game. Follow the fun. It's probably not what you had intended the game to be.
The middle
Throughout the middle of the design, be relaxed. Do not have a picture in mind, of what you want the game to be (as said above). Don't have a time limit. When you reach the middle, just leisurely play the game. Get different opinions. This is where I bring in new playtesters. I want all the different input I can get.
Throughout the middle, the game just gets better with each playtest, and eventually becomes something. Each playtest, I just fix the lowest-hanging fruit. I go over the game's systems and sub-systems one at a time, and the game starts to get good. Once I've gone over everything a couple of times, it all starts to settle down.
Then, the end of development can begin. There's no point (other than the first iteration of the game), where things "click". The game just gets incrementally better.
For some people, the game is finished at this point. For me, it's only a fraction of the way to complete.