I like to stick with rules. Partly that's because I'm not a game designer and I don't like to imagine I know better. Partly it's because I have what might be excessive faith in game designers. Partly it's because I like finding creative ways for the rules to make sense.
A Voice in the Dark:
I also didn't allow Improved Evasion/ Evasion to work when the thief rogue was at the center of a fireball, with absolutely nowhere to hide. Rare situation I know, but they did happen on occasion. (Note I did allow the normal save, and only because they were the thief Rogue).
Good case in point. I used to see it this way. Heck, I had trouble seeing why even a normal saving throw should decrease the damage. It's a ball of fire, right?
Well, then I got to thinking that instead of the rules conflicting with what I imagined was going on, I should let the rules inform what's going on. If I take it as true that a saving throw based on one's reflexes or some other trait can reduce the damage from a fireball and that certain people can avoid damage entirely, then something is going on other than one of those big, orange action-movie explosions. Heck, since the rules make no mention of a blast wave or catching anything on fire, I already shouldn't have been thinking of it like that.
(Or, by that same token, "damage" doesn't mean what I had been taking it to mean. But let's deal with one thing at a time.)
So, whatever the spell was, it was something that was, to some degree, avoidable, even without cover, even if one were standing in the middle of it. What does that thing look like? Well, I don't necessarily know
and I don't necessarily
need to know what it looks like or any of its other properties. What I
can do, what everyone at the table can do, is know the rules of it, and not interpret it in ways that run counter to those rules. Anyone who chooses to do so is inviting conflict between their personal interpretation and how the rules work.
I could go on, and I'm happy to discuss what I mean further, but I wanted to make another point.
I realized at one point that the rules are not the laws of physics of the game world. If you're familiar with the 4th Edition of D&D, that's the game that made me realize this, by presenting multiple ways of handling things. For instance, there were multiple ways to handle monsters: regular monsters, minions (who did static amounts of damage, were dropped by a single hit, and were never affected by a "miss), and by skill challenges.
This blew my mind, and I saw that I could extend this to other aspects of the game. A party sneaking through a castle didn't necessarily have to roll to hit and see how much damage they did when taking out a guard from behind. Sure, there are rules for gaining surprise (higher chance to hit), and for assassinating people (sneak attack, coup de grace and like abilities) and for "unimportant" enemies to get dropped quickly (the aforementioned minion rules), but none of those exactly simulate the situation we've all seen of an infiltrator sneaking up on a guard and taking them out. So - I realized - if I want someone to sneak up on a guard and take them out, and I feel the conditions for that have been met, then it happens, possibly without a roll to hit, or a roll for damage or other things that the rules make available.
At the same time, I completely understand just wanting to see what the rules allow, and based on my example with the fireball I could also look at it from the perspective of interpretation. But whereas the issue the fireball is (as I see it) one of mapping a specific event to a collection of rules, the issue with taking out a guard is one of having rules that don't adequately simulate a specific event.
Which is fine. I've come to realize that rulesystems, even general ones, are designed on certain assumptions about what the game will involve. Dungeons & Dragons, for instance, is best at simulating games in which characters are in close underground quarters, engaged in combat and exploration. Rules for many, many more situations have been devised, but when those rules are put on top of the initial dungeon-exploration rules (rather than, say, being an entirely new set of rules, for, say mass combat), cracks start to show. The rules are a model, and models make compromises and assumptions that make them work less well for certain purposes.
My usual disclaimer: none of this is the last work, I just wanted to offer some other perspectives. I welcome questions and comments. This stuff fascinates me.