There are two competing ideas at play in RPGs, in my experience:
- Rolling is fun. It is generally entertaining to roll dice and see what happens.
- Excessive rolling is tedious. It sucks do roll a lot for little effect.
So when I play games and when I write rules, I generally try to cut down on the number of rolls.
- If a roll doesn't have any risk or significantly exciting drama to it, I'd rather cut it.
- If a roll doesn't determine anything of importance, I'd rather cut it.
- Conversely, if a roll adds drama and excitement, it's worth adding or keeping.
Out of that approach, I have some guidelines I keep in my head when I'm playing or writing:
Don't make me roll to see if I roll.
Generally, if I have to roll dice to see if I get to roll dice, it's less fun than just rolling once. That kind of doubled roll makes me tired all over just reading about it.
So cut down the number of rolls by making a smaller number of big rolls.
Now, sometimes this makes sense. Look at the GURPS combat cycle:
A) Attacker rolls. If the attacker hits, go to B.
B) Defender rolls to defend. If the defender fails, go to C.
C) Attacker rolls damage.
That's an exciting sequence of rolls. Each is pregnant with possibility and tension. It's totally valid to break off the effect roll from the success roll. You don't need to do that - Rolemaster ties the original effect roll to the success roll (although it does add a secondary, and really fun, critical effect roll to most successful hits).
But I came up with things like Flawless Fast-draw and Flawless Nocking in order to, quite simply, because there isn't a lot of drama or excitement in rolling to see if I get to keep rolling.
Multiple rolls to do one thing is only okay if each part makes sense as a discrete unit.
GURPS has a few of these. For example, there is a Will roll to see if you can get your spell roll if your concentration is affected by injury or defending or whatnot, in GURPS Magic. I'm not a fan, although I haven't come up with a seamless and smooth way to replace it, so for now it says.
It's better to influence a roll than roll twice to succeed.
I love things like the GURPS Complementary Skill Rolls. Roll success or failure to see if you get a bonus to another roll. It's especially cool if I get to roll to see if you get a bonus (or a penalty!) from my help. In that kind of situation you double down on the tension without doubling down on the rolling just to see if you succeed. You get more people invested in the process at little cost.
Combine effects where possible.
If you can merge a table so you roll once and look across it for combined results instead of rolling on Table A then B then C, great. If you have a one-shot roll for wandering monsters that combines "does one come?" with "what comes?" that's better than two rolls.
It's not always easy to do this, but if I can, I will do it.
Second chances are okay.
If I roll and succeed, great. If I rolled and failed, a roll to mitigate failure is fun. Missed your defense roll? Roll to see if you're stunned by the blow. Failed that climbing rol? Make a roll to land well.
In short, you want rolls to be an inverted V. One roll expanding to lots of effects, so roll < effect. If you have a small effect for lots of rolls (effect < roll) it's probably too much rolling to get the job done.