| Ring_of_Gyges |
Suppose two people get in a tug of war, the obvious way to do it is an opposed strength check (I make one, you make one, whoever gets the higher result win).
Suppose two people have a dance contest, the obvious way to do it is an opposed Perform (Dance) check.
Initiative checks work this way. I'm going to call all these things "quick contests". Both creatures roll a d20 + something, high result wins.
The trouble I have with the math is that the variation on a d20 is huge in comparison to the variation created by skill disparities. A roll of 8 is low but unremarkable, a roll of 13 is high but equally unremarkable, the difference between them is 5, i.e. enough to wipe out the difference between a Str of 10 and a Str of 18.
If I (lets say Str 10) get in a tug of war with Dwayne Johnson (lets say Str 18) I should lose almost every time, but I don't. The d20 is super random. Dwayne is only beating me about 66% of the time, while my Str 10 equals are beating me 50% of the time. Hell, a Cloud Giant (Str 35) is only beating me 90% of the time. 90% sounds like a lot, but it is a Strength 35 supernatural creature that stands 25 feet tall, I shouldn't ever be beating it.
A +5 difference in an attribute should be huge, its the difference between an average person and a peak human. It just gets swallowed by the randomness of the d20 though.
Is there some graceful way to make skill and talent more important and luck less important. I roll a 17, the Cloud Giant rolls a 4, and I win the tug of war. What happened? He slipped on a banana peel and lost his footing? He got a muscle cramp at just the right moment? In d20 games I often feel like I'm living in a world of banana peels, results are way more random than I'd like in a way that breaks immersion.
One option is to take 10. I get a 10 in the tug of war, Dwayne always gets a 14 and beats me, the giant always gets a 22 and beats Dwayne. That isn't ideal though, since a Strength 12 then beats a Strength 11 100% of the time (which is too little randomness).
Another option is to discourage attribute tests. A dance contest or a tug of war or a research race or whatever isn't one d20 roll it's five. Over the course of those five d20 rolls the randomness recedes and skill will prevail. That's an option, but I don't often see GMs actually do that, "The Orc is pushing open the door, roll a Strength check to keep it closed" is way more common.
Another option is to not use a d20. 3d6 gives a similar range (3-18) but is much more heavily clustered in the 8-12 range. I played in a game at Paizocon which used 3d6 for skills outside of combat and thought it worked well. Our skills much more consistently did what they typically averagely would do.
Another option is to use higher bonuses. A peak human doesn't have a +4 or 5, he has a +8 or 10. The Giant doesn't have a +12, he has a +24. That makes normal human variation (i.e. the difference between an average person and a strong person) more meaningful, but probably introduces other problems.
Anyone else see the same problem? Anyone got a graceful solution?