I just finished one myself...
1. I set the number of talents, powers, and monsters at 20. I figure that's a good number as any to get a couple adventures of material if not more. The sample adventure though had more monsters in it.
2. I wanted the maximum number of skills to be 12, but ended up with 9.
3. I wanted the maximum number of professions to be 7, but ended up with 5.
4. 1 sample adventure, something simple (explore a tomb). Some hooks to go with it.
5. At least 3 campaign secrets so that GM's could write or do their own story arc for discovery. Once the campaign secrets have been discovered and resolved, then I actually consider the campaign (and the game in general) finished.
Actually that sounds pretty cool - simple, but with some thought put in. I'll check that out sometime.
Regarding advice, I'd say you should have a clear vision and stay true to it.
You know Jhaelen I considered saying this, but I think some people have the reverse problem of "not listening to their game."
One of the games I made is an ancient world rpg, with a basis in historical and geographic accuracy. But the Dark Ages part never seemed to shine - I think it's because I stayed
too true to the source material rather than being a bit looser and putting in fantastic elements and fun stuff as I had with the other settings when the historical record (and my own knowledge) was spotty. One of these days I'll see if I can't improve it by loosening it up.
In most cases, though, I do agree with you that it's better to stay true to the original design.
Avoid oddball dice/game mechanics that are only there for the sake of being "unlike any other resolution system"
Your unique game mechanic should solve a problem that other mechanics don't. Otherwise, you are just being different for the same of being different. For instance, using Jenga in the Dread RPG was a brilliant way to induce tension for a horror-themed game. using 20D27 to resolve skill checks because nobody else does is NOT a good decision, when you consider that nobody can find a d27, let alone 20 of them to play your game, and it's really hard adding the result up.
Thank you. Jeez, I've had so many discussions with people about constructed languages, or novels, or rpgs, who insist that there's no such thing as bad design. When you get good at making these things, you find out that there are really some features you can put in that are just plain bad. Definitely my early games had no reason to exist at all.
Still, I wouldn't worry too much about this as a game designer. Mostly the crap will surface when you listen to your playtesters - if you actually listen to them, rather than dismissing their appraisal of your game on the grounds that they just didn't grasp your artistic vision.
(Yes, some playtesters really will look at gold and swear it's lead. But if you keep showing the same thing to different people and they all say they don't like it, chances are it really is bad.)
once I've written the part that I got inspired about everything sort of collapse when I realise I still have to put together some equipment tables or monster rules or other stuff, which I am generally uninterested in but still need in order to make a game playable.
You'd be surprised how little is necessary for a complete equipment table. Usually with equipment, less is more. Heck, send me a PM, meatboy; I'll design an equipment table for you.