1. Does it match the games focus? The core experience of Fallen Enchantress is building, leveling your units, leveling your cities, designing units, building things. Features and ideas that follow this progression are ideal to building a complete theme and feeling for the game. Much like how lighting, music, set design, etc, all work together in a movie to reach a common goal all of the game mechanics should have a common feel that cause the games theme to resonate with the player. (this is why cities having enchantment slots and city types worked so well for Fallen Enchantress)
2. Does it solve a problem? The tendency of designers, including myself, is to create too much. So once I'm beyond the base core of the game mechanics I have to force myself to ask if new ideas and mechanics solve any existing problems. If not, why add it?
3. Does it solve multiple problems? It's usually easy to solve a specific problem. But as I mentioned above, in strategy games every piece affects every other piece. If all an idea does is fix one problem, then it's probably not worth the effort and I usually hold out waiting for a solution that fixes multiple issues at once.
4. Does it have drool factor? (ie: if I tell a player about it will they be excited to try it). I'm less eager to spend development time on things that players won't be excited to try (or as a business guy would say, "doesn't sell copies"). Sometimes we need to spend time on the unexciting parts of the game design, cleaning up screens, polishing graphics, improving performance, fixing bugs, or adding fundamental tools which support gameplay but aren't sexy (I don't think anyone checed out FE and thought "A game with a tax slider?!? I love tax sliders! I'm buying this!").
5. How difficult is it to implement? Now the producer part takes over. Even given all the above the implementation cost could be too high. I love the random maps in Fallen Enchantress, but I don't know if it was worth the considerable implementation time it took to create them. The game is definitely better for them, but we could have created 100 hand crafted maps in a fraction of the time and had all that extra time for other features.
6. How will we provide feedback? A lot of gameplay is feedback. How will the players interact with this mechanic? Will they enjoy it? A lot of mechanics die because there isn't a good way to show the player they are doing it right or wrong.
For the -25% reduction in attack and defense, via Derek's own criteria for assessing ideas:
1. Maybe, maybe not.
2. Sort of, maybe not. Creates other problems.
3. No, creates problems.
4. Definitely not.
5. Easy.
6. No, I don't think gamers will enjoy it.
For increasing armor of garrison troops:
1. Yes.
2. Yes.
3. Yes, solves at least two different problems, probably a lot more.
4. Well, it does make the game more interesting and balanced, because people can feel comfortable knowing that cities of 100s of people won't fall to one Ignys or only a moderate hero. Maybe not sexy, but definitely valuable gameplay enhancement.
5. Easy.
6. Yes, I think they will.
Derek, these are your own criteria, so it would be great if you could re-evaluate the -25% attack/defense idea! We all want the game to be the best it can be, and I think that this idea does not pass the criteria quoted from your post above, whereas the armor idea solves the problem, doesn't create other problems, and even solves other problems currently with the game. 
Yours sincerely,
Steven.
