Evan,
Ah, you exapmle definately helped to illustrate the differences in own gaming styles. IF I were running your adventure I'd have no compunctions about players being killed by ordinary guards. While that result does it does have a negative impact on the game (killing off PCs usually does), so does the other method. I find the sense of danger and risk neceassary to keep the game interesting drops off if the GM is fudging things. Or, to be more precise, if the GM is known to be fudging things. That's what I was referring to by working "under the hood". I'd probably set the situation up to make the guard killing the PCs very unlikely, and trust to the PCs (and thier courage points) for the proper outcome. Likewise, I'd run things about the same as you if the group decides to do something suicidal.
What I wouldn't do is force them towards a specfic resolsutuon becuase of the story. For instance, if the players came up with another way to solve the adventure, I'd allow it as long as it would work and be approriate (i.e. not violate the Prime Directive or some such).
BTW, I agree with you that if you hit a good middleground rule loopholes become less important. In fact, if a game is running well, the rules become transparant and completely unimportant. I've run and played a few advenutres where the group enjoyed a great adventure in a system that was pretty much junk.
Also, how well the rules work depend somewhat on what you are tying to do with them. All RPGs are designed for certain things, and tend to "break down" as you move out of the focus for which it was designed. For example, ICON combat rules don't work well for running an african safari (the elephants are bullet proof). Not a problem since most Trek advenutres don't invlove big game hunting. But, if a GM has an adventure where the PCs do get into such a situation, it helps to be aware of things so he won't be suprised.