In starting to break down all the non-combat bits mentioned previously, I started with alertness. I see the following uses:
- Determining encounter distance
- Determining how quickly someone wakes up at night
- Finding hidden things
- Tracking a creature back to it's lair
- Tracking a fleeing creature
Now I just wrote up a bunch of detailed tables with all sorts of modifiers for these various things. And now I'm thinking, woa, that's way too much for such a simple mechanic.
One problem I see: on the encounter distance, I provide modifiers for various terrain types. Now the PCs probably have limited ability to strategically affect the terrain they're in, other than from a gross standpoint (the next adventure is in the marsh). Modifiers for night are probably ok (even though the players don't get a choice as to whether a random encounter is at night or not - though my thought is actually most random encounters that approach them would do so at night, if you were a bunch of orcs thinking of attacking travellers, would you do so in broad daylight when all their guards are alert, or a night when 2/3 of the guards are sleeping? Especially when you see in the dark better than they do). So I'm thinking of throwing out any modifier based on terrain.
There are lots of modifiers for spells, but I'm thinking perhaps those could go in the spell descriptions (afterall, the combat rules don't talk about all the spells that modify to-hit probabilities). The spells along with character generation/advancement provide a variety of tactical and strategic choices which should make this an effective rule.
Waking up - in one sense, I'm a jerk and don't let people wake up instantly. On the other hand, I see some interesting choices introduced. What I'd like is a systematic way of handling it that allows the players to see who they need to make an effort to wake up, and who will wake up on their own. In the past, we had some people sleep through several rounds, which does start to get ridiculous. Hmm, and the chart I worked up doesn't look very good (the worst PC has like a 50-50 chance of sleeping at least one round after being kicked - or struck by an enemy). One thought is to have a defined "what is necessary to wake this person up" depending on their alertness (so someone with a poor alertness needs to be kicked). If you do the defined thing, they will wake up the next round. And then perhaps a one lesser step where they wake up in two rounds. Anything less and they must roll (and could sleep through it - I don't feel too bad that someone might sleep through a fight going on say more than 20 meters away from them). I need to think about this, but I do want sleeping to be a real penalty (I stopped doing night encounters in D20 partly because it really didn't seem to be much of a penalty at all if I ran the rules as written).
Tracking: I was splitting tracking into two types. One is that a successeful tracking roll produces an "optional" result (like extra treasure), the other is when the GM intends the quarry to be found (tracking the orc scouts back to their cave which is the object of the adventure...). In the "intended to be found" result, a good tracking roll would reduce the encounter distance (you sneak up on the cave, or they spot you 100 meters out) and/or find a back door or other "best" approach. The "optional" case is a pure success/failure. Again, I put in all sorts of modifiers. I'm inclined to replace the modifiers with standard DCs (wolf lairs are DC 20, orc lairs are DC 10, dragon lairs are DC 30). Again, modifiers for weather and such aren't something the PCs can do much about.
Searching: here I was thinking of set DCs for "optional" things, differing time for "expected" things (time to find something plays nicely into the system since spells will wear off and such, so this is a useful result). I guess optional things could actually be rolls on the time chart, with a modifier for difficulty, and a minimum result needed (the secret treasure compartment is -10 to your check, you must roll at least a +0 result, which results in an hour of searching). The PCs of course can always choose to cut the search short. Ask them what the maximum time they want to search is - if their result indicates a longer time than that, they fail and spend that much time. It might also be fair on a really terrible roll for them to only spend 1/2 the time, and realize there's no hope (you start searching the haystack, after 5 minutes, you realize there's no way you'll search the whole stack in 10 minutes [the roll was going to require an hour of searching]).
The big key here is that these are all at most a single roll per PC. Searches might be nice to figure a way to combine checks so the whole group makes one roll (the logarythmic scale for adding grapple abilities together might work, even the PCs with an 8 alertness will add +1 or +2 to the total). Wake up definitely needs to be per person Encounter distance, which gives a possibility for the really high alertness PC who is sleeping to notice - and thus wake up w/o a wakeup check, is probably also best rolled by each PC (that also gives the best benefit to multiple people on watch).
The most important ones to the core story of course are the encounter distance and wake up checks. Tracking an "intended to be found" encounter is also part of the core story. The other checks are mostly for extra stuff, though I think the search for hidden treasure does play well into the kill things and take their treasure core story. The trick is to make sure every roll really is helping focus the intended game. Otherwise the answer should be "yes" (though I think "no" is also reasonable if it is totally unreasonable - for example, in a conflict res game, players can say "no" to unreasonable stakes - though something to consider is that when you are negotiating unreasonable stakes into reasonable stakes, I think that's actually a social contract negotiation ["Let's talk about these stakes I want to become king, that's not the game I was interested in playing, how do we come to agreement here?"] - once the social contract is successefully re-negotiated, we step back into the game, and say yes or roll the dice).
One thing I like about Cold Iron is how it systemizes modifiers, so for example, most things that penalize spell success break down into a -3 or -6 penalty based on degree of interruption of concentration. Lots of combat penalties amount to a -4 defence for "really bad position". Or -6 for attacking when not really ready (drawing a weapon, or aborting a full defense to swing at someone who decides to grapple you). I think the alertness system needs to have a similar level of complexity (and probably should try and use the same modifiers, though being an attribute check, the scale is sort of different).
Hmm, I also need to systemize some of the other things I use alerness for (for example, realizing your non-magical sword isn't actually damaging the creature, or noticing whether the creature has been affected by a spell or not).