1) Start with a concept. Anything will do, it's not really going to matter. The standard form of "just like X except with Y" is perfectly fine.
2) Believe that it's a real game. It already exists, in a sense; games are mathematical objects, they have independent existence in the Platonic world of forms. Or something. Whatever it takes to convince you that it will work.
3) Start with the smallest thing that the player does regularly in the game. If they're mostly moving a character around, then that. The core interaction. Code it.
4) Spend lots of time on graphics, and on getting the feel of the controls right. These are really important for being able to trick yourself that this could plausibly be an Actual Game.
5) Play with it. Lots. Pretend the rest of the game is there. Yay, you made a game! (Now you just need to finish it.)
6) Drop your original concept. There's not enough time to implement it, and it's probably stupid anyway, what were you thinking? Just forget it. What you've done already can work without it.
7) Okay that's probably left some threads hanging: tie all of them together. You had this big idea but now that you don't have time for it there's a gap there, something else needs to fill its role. Just join everything up, it'll work out somehow. If there are two characters that aren't properly fleshed out - just make them the same person. If you've dropped your elaborate crafting system but the player still needs to acquire items somehow, and you've dropped your complex AI controller but you still need to introduce enemies somehow - just make the items be the enemies. Does that even make sense? It doesn't matter.
8) Keep playing the game. This is important, way more important than programming it. You need to get a feel for what it wants to be.
9) Realise it's not going to work, and throw it all away to start again. It's stupid, it's not interesting at all, and even at its newly reduced scope there's no way you're going to finish it, you were just deceiving yourself all along. You've probably still got time to make something really small... goto 1. (NOTE: skip this step in the event that your game is actually any good.)
10) Uh if you reached this then you have something interesting, keep working on it until it's done I guess? Don't worry too much about the deadline, it's better to make something good than force it into some arbitrary constraint. Neat.