I explain most of the concept and origin of the game on the challenge page, but here are a couple additional thoughts. I really wasn't able to scope it down enough to fully explore feature creep. I still get ideas from the original prototype that fit in the parameters, and from where I stand now, I think there are three possibilities.
First, you tighten your scope with strict rules and limits, even something as severe as you can only output a 1 or a 0, and you will still be able to feature creep yourself into crunch time. Many humans are amazingly creative, and their minds are built around coming up with ideas and suggestions in the face of limits. The only way to stifle feature creep, is to kill creativity. This is what most time-limited projects are forced to do, sometimes badly.
Second, you allow feature creep, and you don't want to kill creativity, so what do you do? You have to find a filter of some kind- a way to say, "it's done." Some of the best games have that filter by utilizing a single, focused creative director who owns the vision and is good at expressing the vision both artistically and technically. Other great games filter by committee agreement- more cumbersome but arguably able to please more people. Other games filter by bug count or feedback ratings or something similar. As a minor note, in this scenario, some of the feature creep suggestions will conflict with each other and decisions will have to be made about which thing to include.
Third, of course, is an infinite project. Perhaps World of Warcraft will exemplify this, if it never dies. Perhaps there are other open source projects that are like this. Perhaps in the future there will be AIs that make this happen as well. Something like this is certainly in the realm of thinking big.
Anyway, enough rambling. Here is the game. Please go check it out and feel free to comment or tweet and let me know what you think!