Yep, the variance is more in the powers. The question is would standardizing the defenses take out any appreciable flavor. With only 4 defenses, I say that making them rigidly standard does take out appreciable flavor, but it doesn't take much to fix this.
With 4 defenses, you've got a very limited number of variations that will be
noticed. You can have a monster sitting at baseline for his level on all defenses, of course. Then you can have:
- Appreciably better at one of the four things.
- Appreciably worse at one of the four things.
- Appreciably better at two of the four things.
- Appreciably worse at two of the four things.
That's it! Appreciably better at three of the four things is the same as, "higher level, but worse at the remaining thing." When I said 80% by the system specified, my thinking is that the two out of four cases won't get noticed much by the players.
Take a dragon, for instance. Dragons are either standard, or they are "Bad Reflex"--most likely. They might be "Good Fort" or "Good AC" for some people. They have thick scales. They are tough. They are often stubborn. So depending upon how you want to do it, you make them higher level with "Bad Reflex" or you make them lower level with "Good Fort" or "Good AC". The only corner case you are missing is if you absolutely want AC and Fort to be appreciably better, Will to be standard, and Reflex to be substandard. The players won't notice that based on defenses alone. If you want them to notice that kind of subtle distinction, you'll need powers that make it clear, anyway.
That said, my idea is merely a simple example. You could just as easily stick to something closer to the current method by defining a handful of standards for each type. If soldiers get one adjustment and skirmishers get another, fine. But once you get, say, around 6 variants that cover most of the defense cases, you really aren't buying yourself much to finish filling out the rest.
In an otherwise standard creature, -1 Reflex and -1 Will is mere noise.
Note that there may be more serious repercussions in the actual results, due to changes in the math. I don't know about that either way--don't even really have a hunch. My claim is merely that the players won't notice the changes, absent serious and sustained investigation into the math.