I thought I’d contribute to Martin’s Blogging for GMs project.
Last night Jennifer and I were talking about calling for die rolls in adventures. I repeated a bit of truth that I’d read on an advice site: don’t roll dice when it doesn’t matter. (I can’t track it down at the moment; it was delinked long ago.)
For example, if the PCs are coming up on a hallway junction, it’s usually not worthwhile to call for listen checks. Why? Well, you’ll generate a result per person, usually with the effect that someone will succeed. (Enough die rolls and someone’ll usually succeed.) And they’ll turn to their companions and say “I heard X”. All of the time checking skills, gathering rolls, etc. boil down to a bit of information everyone winds up with anyway, which is a waste of time.
That isn’t to say “never make listen checks”. If you’re going to make the check, the check needs to have meaning. Maybe only the people who beat a specific DC can act the first round… because you’re actually hearing their attack and don’t have time to react to someone else’s warning.
Move Silently is strange too. If you have Sneaky McSneakerson trying to get by, the best defense is hiring more guards, not better guards. And if you have the PC make multiple sneak rolls and use the worst as the target of the listen check, Sneaky’s in really in bad shape.
Having 5 people listen (each getting to roll listen) is more reliable than hiring a character with an extra 10 ranks in a skill. (5 people have a 67% chance of having at least one character roll 17 or better. They also have a 97% chance of getting at least one 11 or better.)
Search is similar; having people roll to find a critical element will result in rerolls or accepting the best roll as “just high enough”– a soft kind of fudging and a waste of time. What do you do when they fail their search but describe taking the proper action to make the thing work? Under what circumstances can you search again?
Solutions? I don’t promise anything, but these might help.
- Roll once (for each side) and add the highest check to the result. This rewards highly skilled people (on both sides), but may mean that each character specializes in a skill and others don’t bother with it. Plus players aren’t rolling the dice, which denies many a source of happiness.
- Jennifer suggests pre-rolling. At the beginning of the session, have everyone roll d20 five times, with the GM marking down the results. Tick off the checks (so you don’t alert the players that there’s something to pay attention to here) and have them roll more for your list periodically.
- Searches are tricky and can get repetitive quickly. I suggest rolling once per room (with teamwork bonuses), and possibly using the checklist to hide the value (to discourage rerolls because the die came up low).
- Setting clear stakes before you roll can keep everyone on board, and alleviate misperceptions like “roll until you fail”.
6 replies on “D&D die rolls: When and why”
The first time I saw the “roll once for everyone”, it was Lee Gold running Lands of Adventure. I really cringed because all that she cared about was the highest skill (so she might just as well have asked “who has the highest skill”).
For searching, we make heavy use of Take 10 and Take 20. Usually it ends up just being the highest searcher, but if the search is easy, we’ll divide the time up (let’s say they take 10, and searching the room will take 10 DC 20 successes. If there are 3 characters with a Search of +10 or better, it will take them 7 rounds to search the room).
I also make use of roll Spot or Listen against some DC to not be surprised (usually a Sneak check). One way to reduce the randomness of sneak checks would be to assume they are always done with Take 10 – and then give circumstance bonuses.
Of course if it’s just a case of does someone hear the noise, well there should be a benefit to having more listeners. Part of the problem is that for D&D, a d20 provides too much randomness for the range of skills often encountered. Of course if the DC is high enough, you will be better off with one person with 10 more ranks than the horde (especially since a 20 doesn’t convert to automatic success). If the horde needs to roll a 20 to succeed, 15 or 16 horde folks will have about the same chance to get one success as 1 +10 guy. Now if the horde only needs a 15, then 4-5 of them will be as good as the +10 guy.
One option is that if only one success matters, you could come up with some scheme where the horde combines into a single check. A homebrew a college friend developed used a combinational scheme for grappling where the individual grapples were combined as if they were logarithmic. So your grapple represented LogX(Y)=G, and I think in specific, LogX(2)=6, so two people with the same grapple G1 had a combined grapple G2 of G1+6 since combining the two grapples is the same as multiplying by 2, so you add LogX(2). Now adding the 3rd guy only added +4 more. So if we used this same scale, we would know that three people were as good as someone with a 10 better skill, and that would be constant no matter what the DC was. Now you just need to decide how many people with +0 listen should someone with +10 in sneak be able to evade 50% of the time, and set your log scale appropriately (so let’s say you want it to take 10 people, well, set LogX(10)=10, or if you wanted it to be 20 people, set LogX(20)=10).
Another way would be to decide that while a few extra ears helps, a crowd doesn’t. So you decide for any given sneak check, only the 4 best listeners get to roll.
Aren’t probabilities so much fun?
Frank
Strangely, my first response was eaten.
You make excellent points, and further examples are very handy, thank you.
I like your college friend’s log scheme, but doubt I’d actually use it in session. (Though I might pre-generate the numbers and put them in the screen as a time saver… hmmm.)
I agree, Take 10 and 20 are solid ways to cut time (especially if combined with the log advantage above…) I’ll mull it over and probably scribble yet more.
My friend’s system included the log chart on the reference charts page (which everyone had to use, because another thing he had was a scheme to use the normal distribution instead of a d20, basically you rolled d10s to generate digits of a number between .0000… and .9999… and looked the number up and read a modifier off the table – if you want to know more, check out my Cold Iron page: http://www.mindspring.com/~ffilz/Gaming/ColdIron. The Game Reference charts have the chart, the Combat rules tell how to use it).
Frank
I’ve been thinking about this; while I’m not aiming for a log chart, I may make an average value chart– useful for converting multiple opposed rolls into a static DC.
I like these timesavers, Scott — particularly the one about searching once per room. Searching just isn’t interesting enough to warrant much more attention than that.
As far as log charts go…shudder. 😉 Unless something like that is completely fluid and easy to use, it makes my eyes glaze over when it crops up in a game.
The log table isn’t too bad:
For multiple people with the same grapple:
2 +6
3 +10
4 +12
5 +14
6 +16
For additional people with varying grapple (take the total so far, and compare the next person):
delta add
+0 6.0
-1 +5.5
-2 +5.0
-3 +4.5
-4 +4.2
-5 +3.9
…
-9 +2.6
The key was that mostly only the bad guys would have multiple people in the same grapple (grapples with multiple people on each side broke up into smaller grapples). And of course the bad guys are almost always clones (especially the ones that grapple in mass, like wolves and ghouls – if you’re being grappled by multiple trolls, you’re dead… and multiple balrogs didn’t grapple – they usually grappled and then detonated a ground zero fireball [which was normally a single target spell, but hit everyone in a grapple, of course balrogs are immune to fire….]).
Something to consider on search if it really belongs in the game. It seems like the primary uses of search are:
– to actually get your reward
– to have the opportunity to avoid a trap, which is just random damage with little opportunity for strategy
– to continue the adventure (the old secret door)
Occaisionally a search check finds something non-critical that makes the adventure easier. But is it really good design to have a significant shortcut depend on a single random roll?
Of course a search check in Donjon (as I understand – I don’t have the game) plays an entirely different role.
I’m going to be expanding on this idea some in my blog (I sort of started a discussion there about D&D’s complexity):
http://welcometofranksworld.blogspot.com/
Frank