I can't stop thinking about different dice resolution systems.  I feel like the "best one" is out there somewhere, and feel frustrated that I don't know what it is.  For example, I prefer target number systems with cumulative dice.  But I also prefer easy-to-follow target numbers, like percentages.  I think in percentages already anyway, so that's very natural to me.  And as a GM, it'd be easy to say guess a percentage of likelihood: "What are the chances you would fail to climb that wall?  Oh... 30% sounds good."  So today I'm noodling how to use cumulative dice to climb towards a percentage number.

I'm thinking factors.  Each one gives you (some) dice.  I'm also wondering if it would be cool or annoying to use differently shaped dice, so maybe:

Skill of 5d10, weapon quality 2d6, blessing 1d4 aiming for a Target Number of 60.

It's just a rough example.  Have you seen this before?  Am I remembering it from somewhere?

What do you think of it as a general idea?

What are you doing in your example? Adding the dice all together? That's a lot of things to add, and not terribly intuitive for me.

Quote from: mcbobbo on March 29, 2022, 03:51:30 PM
I can't stop thinking about different dice resolution systems.  I feel like the "best one" is out there somewhere, and feel frustrated that I don't know what it is.  For example, I prefer target number systems with cumulative dice.  But I also prefer easy-to-follow target numbers, like percentages.  I think in percentages already anyway, so that's very natural to me.  And as a GM, it'd be easy to say guess a percentage of likelihood: "What are the chances you would fail to climb that wall?  Oh... 30% sounds good."  So today I'm noodling how to use cumulative dice to climb towards a percentage number.

I'm thinking factors.  Each one gives you (some) dice.  I'm also wondering if it would be cool or annoying to use differently shaped dice, so maybe:

Skill of 5d10, weapon quality 2d6, blessing 1d4 aiming for a Target Number of 60.

It's just a rough example.  Have you seen this before?  Am I remembering it from somewhere?

What do you think of it as a general idea?

Roll over or roll under? Because with the dice you picked and the target number, that's either nigh impossible to succeed or nigh impossible to fail.

Way too many dice to add up in that example for things to resolve quickly unless you're already really good at that type of addition.

A related factor is that because you're using multiple dice its NOT a straight percentage chance you're rolling for in your example. The average on 5d10+2d6+1d4 is just 37 and the odds of achieving a 60 or better result are virtually nonexistent.

The big things to consider are...

A) do you want margins of success or not? If yes, then you'll want multiple dice to get a bell curve distribution of results, but once you're at 3 total dice the degree of closeness to a bell curve is generally not worth the extra counting time. If not, keep it simple with one die (or percentage dice) + modifiers.

Also worth noting is that any task which requires multiple action rolls to succeed is automatically a bell curve of sorts.

B) How often are people rolling? Percentiles are great if you have lots of rolls because there will be enough rolls for 1-2% to matter. But if players are only making a dozen rolls per session, that degree of precision may not actually come up enough to really matter and you're better off using smaller numbers (d20 being used so much isn't exactly an accident... it's basically 5% increments of probability which is at the upper limit of what most people can estimate (leaving aside the Dunning-Kruger effect).

Honestly, a simple percentile roll under skill level system with bonuses and penalties is probably what you'd actually want to look at if you really want a percentile based system.

Quote from: mcbobbo on March 29, 2022, 03:51:30 PM
I can't stop thinking about different dice resolution systems.  I feel like the "best one" is out there somewhere, and feel frustrated that I don't know what it is...

Resolution mechanics don't exist in a vacuum.  They are part of the system.  A good resolution mechanic is in part one that fits the overall system well.  So by definition, there can't be a "best one" out there. 

Now, narrow done what are the goals, restrictions, etc. of the overall system first.  Then the list of resolution mechanics that will be a better fit will get short in a hurry.  If you go the other way, and start with a mechanic you really like, then that necessarily constrains the list of system aspects that will fit well around it.

I vaguely think the idea of adding up the total of a pool of different types of dice may have been done in Earthdawn, and I know there have been systems which used a Stat die plus a Skill die with each larger grade of die representing a step in ability. I don't think I've ever seen a pool with as many dice as you propose, though.

Some points I would note are:

- Once you get past three or four dice on a regular basis, counting successes is going to be measurably quicker than adding all the dice up for a total. Even with counting successes, keeping the physical number of dice rolled at a time to 10 or fewer is probably a good idea.

- If you want to use different types of die in the same pool, I'd suggest using a set target number of 4, as this is the only way a pool component measured in D4s can contribute.  Success is then measured in the number of "hits" (4 or over) the dice produce.

- You are unlikely to be able to get an immediate, workably accurate estimate of a simple percentage chance this way, as the cumulative probabilities for dice of different types rolled together is tricky for most people to calculate on the fly.  You may want to figure in advance some "benchmarks" and let players wing from there based on intuition -- for example, with a set TN 4, a pool of 5d6 requiring 3 hits is exactly 50-50% succeed or fail odds.

(I personally like dice pool systems precisely because they make final numerical probability opaque in this way -- to me, thinking in terms of "a 73% chance of success" is immersion-breaking -- but it sounds like you prefer systems where knowing your final chance is a little easier.)

Quote from: Thondor on March 29, 2022, 04:07:42 PM
What are you doing in your example? Adding the dice all together? That's a lot of things to add, and not terribly intuitive for me.

That's the thought, yes.  I usually use a computer to play, but I did see that possible problem.

OTOH, it could possibly make the dice goblins happy to roll so many.

Quote from: migo on March 29, 2022, 04:09:51 PM
Roll over or roll under? Because with the dice you picked and the target number, that's either nigh impossible to succeed or nigh impossible to fail.

Ignore the example's TN, sorry.

Roll over.  Cumulative towards a target, so more dice is always better, if only by +1.

Quote from: Chris24601 on March 29, 2022, 04:20:48 PM
The big things to consider are...

A) do you want margins of success or not? If yes, then you'll want multiple dice to get a bell curve distribution of results, but once you're at 3 total dice the degree of closeness to a bell curve is generally not worth the extra counting time. If not, keep it simple with one die (or percentage dice) + modifiers.

Also worth noting is that any task which requires multiple action rolls to succeed is automatically a bell curve of sorts.

B) How often are people rolling? Percentiles are great if you have lots of rolls because there will be enough rolls for 1-2% to matter. But if players are only making a dozen rolls per session, that degree of precision may not actually come up enough to really matter and you're better off using smaller numbers (d20 being used so much isn't exactly an accident... it's basically 5% increments of probability which is at the upper limit of what most people can estimate (leaving aside the Dunning-Kruger effect).

Honestly, a simple percentile roll under skill level system with bonuses and penalties is probably what you'd actually want to look at if you really want a percentile based system.

A:  Not.  I don't like 'number of success' systems, just in general.  Straight pass or fail for the core mechanic would be my preference.

B:  Assume reasonably often.  Not necessarily a pure OSR approach, but neither a completely gamist one.  Somewhere in the middle.

And certainly a simple d% is easier, but there's no mystery to it, and I find it kind of boring.

Looking to spice things up by obfuscating from the brain what the dice might do.

Quote from: Steven Mitchell on March 29, 2022, 04:50:54 PM
Resolution mechanics don't exist in a vacuum.  They are part of the system.  A good resolution mechanic is in part one that fits the overall system well.  So by definition, there can't be a "best one" out there. 

Now, narrow done what are the goals, restrictions, etc. of the overall system first.  Then the list of resolution mechanics that will be a better fit will get short in a hurry.  If you go the other way, and start with a mechanic you really like, then that necessarily constrains the list of system aspects that will fit well around it.

I want a dice mechanism that determines success or failure of an attempted task.

Quote from: mcbobbo on March 29, 2022, 05:19:13 PM
Looking to spice things up by obfuscating from the brain what the dice might do.

Apologies for misreading your original post; I had thought you were interested in systems that made the final chances obvious the way basic roll-under-% systems do.  My bad for not paying closer attention.

It also occurs to me that this thread might belong more in the "Design, Development and Gameplay" forum.

Quote from: Stephen Tannhauser on March 29, 2022, 05:12:53 PM
I vaguely think the idea of adding up the total of a pool of different types of dice may have been done in Earthdawn, and I know there have been systems which used a Stat die plus a Skill die with each larger grade of die representing a step in ability. I don't think I've ever seen a pool with as many dice as you propose, though.

Some points I would note are:

- Once you get past three or four dice on a regular basis, counting successes is going to be measurably quicker than adding all the dice up for a total. Even with counting successes, keeping the physical number of dice rolled at a time to 10 or fewer is probably a good idea.

- If you want to use different types of die in the same pool, I'd suggest using a set target number of 4, as this is the only way a pool component measured in D4s can contribute.  Success is then measured in the number of "hits" (4 or over) the dice produce.

- You are unlikely to be able to get an immediate, workably accurate estimate of a simple percentage chance this way, as the cumulative probabilities for dice of different types rolled together is tricky for most people to calculate on the fly.  You may want to figure in advance some "benchmarks" and let players wing from there based on intuition -- for example, with a set TN 4, a pool of 5d6 requiring 3 hits is exactly 50-50% succeed or fail odds.

(I personally like dice pool systems precisely because they make final numerical probability opaque in this way -- to me, thinking in terms of "a 73% chance of success" is immersion-breaking -- but it sounds like you prefer systems where knowing your final chance is a little easier.)

Yes, could be Earthdawn.  It's definitely been a minute since I ran that one...

Counting successes is anathema to me, because dice don't care about probability curves.  Not going to rehash that argument, so feel free to disagree, but adding dice to a number of success should not be assumed to increase the occurrence of a success.  The pretty curves will imply that you'll HAVE to roll a success from a certain number of tries, but ultimately it's not true.  Plus, and this is just my experience of course, failures tend to happen at the worst possible times under these systems.  "Easy" rolls do not exist when zero successes is a potential result.

Not trying to cut off what may be an interesting discussion, but that's really all I'm going to say about it here.

Sum up to say, I prefer cumulative systems where you're adding the dice towards a result because more dice is always better.

Non-100-based TNs are doable, but not as easy to estimate.  A d20 is after all relatable to 100 by 5, but few people look at it that way.  People equate 5e's advantage to +5, not +25%, for example.

Estimating the likelihood of success of the roll is not a goal I have in mind.  The opposite, actually - I want it to feel like a risk, even when a lot of dice are involved.

Quote from: Stephen Tannhauser on March 29, 2022, 05:27:13 PM
Apologies for misreading your original post; I had thought you were interested in systems that made the final chances obvious the way basic roll-under-% systems do.  My bad for not paying closer attention.

It also occurs to me that this thread might belong more in the "Design, Development and Gameplay" forum.

No worries at all.  It's been a minute since I've posted here.  If moving is a thing that can be done, I'm of course open to that happening.

Quote from: mcbobbo on March 29, 2022, 05:27:59 PMEstimating the likelihood of success of the roll is not a goal I have in mind.  The opposite, actually - I want it to feel like a risk, even when a lot of dice are involved.

That makes sense. Okay, assuming the goal is for the dice to total over a number between 1 and 100, I then have some further thoughts:

- Every die will add half its maximum value (+0.5) to the average outcome of any pool it's part of.  You can therefore think of every d4 as being "worth" 2.5, every d6 as being "worth" 3.5, etc.  To get a feel for the Difficulty TN of various tasks, put together the typical die pool of an "average" person, figure out its average outcome (e.g. 3d6 + 2d4 = 10.5 + 5 = c. 15-16).

- Then decide whether the TN of an "average" difficulty task should be at the point where it's always a 50-50 risk or lower (i.e. anything you actually roll for is by definition difficult and risky), or whether it's at the point where you're more likely to succeed than fail (i.e. in real life most of us succeed at the vast majority of our professional uses of our skills, or we wouldn't still have jobs). Place it as appropriate compared to that baseline average outcome, over or under.

- Use that "midpoint" difficulty as how to assess difficulties going above that.

One thing as noted above is that going by this analysis, getting totals as high even as 20 on a reliable (i.e. more than half the time) basis will require pools of 4d10 or more, or the equivalent. Getting any total of 50 or higher half the time or more requires 10d10 or 5d20. You may have to assess how much character development is required to reach that point.

Quote from: mcbobbo on March 29, 2022, 03:51:30 PM
I can't stop thinking about different dice resolution systems.  I feel like the "best one" is out there somewhere, and feel frustrated that I don't know what it is.  For example, I prefer target number systems with cumulative dice.  But I also prefer easy-to-follow target numbers, like percentages.  I think in percentages already anyway, so that's very natural to me.  And as a GM, it'd be easy to say guess a percentage of likelihood: "What are the chances you would fail to climb that wall?  Oh... 30% sounds good."  So today I'm noodling how to use cumulative dice to climb towards a percentage number.

I'm thinking factors.  Each one gives you (some) dice.  I'm also wondering if it would be cool or annoying to use differently shaped dice, so maybe:

Skill of 5d10, weapon quality 2d6, blessing 1d4 aiming for a Target Number of 60.

It's just a rough example.  Have you seen this before?  Am I remembering it from somewhere?

What do you think of it as a general idea?

I'm not a fan at all of percentage rolls. But if I were, then how would you handle degrees of failure/success? Or would it just be pass/fail?