Hopefully this is the first of a regular multi-part series. This will be pretty heavy stuff, so feel free to tune it out. The idea is to chronicle design decisions behind writing an RPG. I've attempted this before, but never really did anything with it.
First principle of the design process is deciding whether or not the system will be tied to a particular setting or not. I'm not out to build a be-all-end-all system like GURPS or Superlink: I'm out to build a system that suits my particular tastes and needs. So it'll be tied to support the tropes and conventions I like in my games and stories. Admittedly this is a wide variety, so it'll probably be a broad spectrum of mechanics anyway.
The second principle of the design process is complexity. I want the system to be fairly easy to remember, so the key here is to devise a core mechanic that is consistently applied across all instances of task resolution. RPGs are not in the same embryonic state they were when D&D 1st Edition was being molded; there is no reason for a wide variety of task resolution mechanics to exist like they did a few decades ago. A consistent mechanic also makes the game easier to learn and easier for the Game Moderator to 'guess' when it comes to weird unforeseen circumstances.
The third thing I have to consider are the dice. I like random chance in my games, so I've already made the decision to have dice in the mechanics. But size matters. Do I want a curve or a flat probability? How big should that curve be? Do I want just one die type to rule them all? Should the dice 'explode'? Is it a dice pool system? There's a lot to consider and they all have an impact on the complexity of the system. As exciting as all that could be, I'm going to stick with a d100+modifiers vs. a fixed target number of 100, with the margin of success/failure determining many of the mechanical effects.
That's probably enough for right now.