PDA

View Full Version : D&D 3.x Other Has anyone tried implementing a dice-pool style change to how skill checks work?



Ephemeral_Being
2015-03-07, 10:55 PM
I've been thinking recently. The combat rules in 3.5 are actually pretty decent, both realism-wise and to use. But using the d20 system to adjudicate skill checks has been frustrating me to no end. You roll on the d20 is a colossal proportion of the total score, especially at low levels. A first level character with a Listen skill of 5 (which is entirely reasonable) has something like a 25% chance to hear a character with a Move Silently score of 15. Which is what you'd see on a level 7ish master thief.

Now, systems that use a dice pool are comparatively excellent at resolving skill checks. You have a number of ranks in a specific skill (say, tumbling), and roll a number of d6 equal to those ranks when you use the skill. A certain range of numbers is a denoted a "success." You need a certain number of successes to do certain things. When you have opposing rolls (Stealth vs.Move Silently), you each roll your skill and compare the number of successes. Some systems use 3-6, others 5-6 as a success.

What I was wondering is, has anyone tried combining the two in 3.5? Use the d20 combat system with dice-pool style skill checks? Without any changes other than replacing the rules behind skill checks? What range is best for determining a "success?" How/where did you set the number of successes for each DC? At 10/15/...35? Should that scale linearly? Or, is there a better way?

I was thinking 5-6 is a success, and a DC of 10 is one success. DC 15 is 4 successes, and it scales linearly. But I haven't had a chance to try this out.

Grod_The_Giant
2015-03-08, 06:23 AM
There's nothing really wrong with the idea, apart from internal consistency, but I think it might require more work then you'd expect. Given the huge range of possible skill modifiers, even at the same level, you'd probably have to rewrite the entire system for how skills work.

As a possible compromise, how about replacing the d20 with 3d6? The range of numbers is almost the same, but the results are much less random, forming a bell curve around 10 rather than an even distribution.


I was thinking 5-6 is a success, and a DC of 10 is one success. DC 15 is 4 successes, and it scales linearly. But I haven't had a chance to try this out.
This being on a d20? Because that seems like way too low a margin. I'd think you'd want a success to be in the top 60-70 percentile, meaning a 12-14 on a d20, with a natural 20 counting as two successes.

johnbragg
2015-03-08, 08:09 AM
There's nothing really wrong with the idea, apart from internal consistency, but I think it might require more work then you'd expect. Given the huge range of possible skill modifiers, even at the same level, you'd probably have to rewrite the entire system for how skills work.

As a possible compromise, how about replacing the d20 with 3d6? The range of numbers is almost the same, but the results are much less random, forming a bell curve around 10 rather than an even distribution.
.

Ninja'd. I'm still noodling an idea for a Rogue fix/revamp based around the idea of letting Rogues (Chancers) roll 3d6 instead of d20 whenever they want to (when they just want to avoid fumbling, or rolling a 4 etc) and spending power points to roll 4d6 or 5d6 or more when they want or need to. (And, if you're giving them a pool of power points, you might as well give them a small selection of low-level spells as SLAs on top of that.)

As skill-monkeys who also see combat, and tend to have a relationship with luck as a law of physics, they get an intuitive sense that their world is in some way an overlapping, interlocking set of probability fields. And, as they grow in power, they figure out how to adjust these probabilities in their favor. To a thief or rogue, only a sucker plays the game by the rules.

Inspired in party by Haley's speech here, elevated to an in-character essay on metaphysics in a D&D world. http://www.giantitp.com/comics/oots0428.html

Ephemeral_Being
2015-03-08, 12:15 PM
There's nothing really wrong with the idea, apart from internal consistency, but I think it might require more work then you'd expect. Given the huge range of possible skill modifiers, even at the same level, you'd probably have to rewrite the entire system for how skills work.

As a possible compromise, how about replacing the d20 with 3d6? The range of numbers is almost the same, but the results are much less random, forming a bell curve around 10 rather than an even distribution.


This being on a d20? Because that seems like way too low a margin. I'd think you'd want a success to be in the top 60-70 percentile, meaning a 12-14 on a d20, with a natural 20 counting as two successes.

I meant 5-6 on a d6. Meaning a 33% chance to get a success. I was literally trying to straight-port the mechanic over.

My friend pointed out something that I hadn't considered. At level 1 (and even past that), you can have zero or even negative ranks in a skill. When the DC to hear something is 10-15, it's RARE that you'll hear it, but not impossible. What I was proposing would make a character with a Wisdom of 10/11 and no ranks in Listen functionally deaf.

I like the idea of 3d6 rolls for skill checks. That's probably a more reasonable distribution. It's something we had previously played around with, but you still ran into the same issue of your actual skill ranks being largely auxiliary to luck until around level 4-5. At which point you can finally have enough skill ranks/equipment/stats to get a decent modifier, and actually manage to beat a first level character with an excellent dice roll. Which was what the primary goal here was.

Last session, I had a first level Rogue attempt to steal from our party's 7th level Ranger. A Spot skill of 18, versus the Sleight of Hand skill of 4. And yet, despite the incredibly low odds, he had his coin purse stolen. Because he rolled really crappy, and I rolled really well. Now, yes. As the DM, I can fudge that. But I don't want to. The mechanics allow for that to happen. And, after listening to him rant about losing 10,000 gp because of luck, I have to agree. That IS stupid. That's really stupid. Hence trying to improve the mechanic. One that makes opposed skill checks primarily dependent on the character, and not luck.

I'm not giving up. There's some way to make them more fair. I just haven't found it yet.

johnbragg
2015-03-08, 01:43 PM
I meant 5-6 on a d6. Meaning a 33% chance to get a success. I was literally trying to straight-port the mechanic over.

My friend pointed out something that I hadn't considered. At level 1 (and even past that), you can have zero or even negative ranks in a skill. When the DC to hear something is 10-15, it's RARE that you'll hear it, but not impossible. What I was proposing would make a character with a Wisdom of 10/11 and no ranks in Listen functionally deaf.

I like the idea of 3d6 rolls for skill checks. That's probably a more reasonable distribution. It's something we had previously played around with, but you still ran into the same issue of your actual skill ranks being largely auxiliary to luck until around level 4-5. At which point you can finally have enough skill ranks/equipment/stats to get a decent modifier, and actually manage to beat a first level character with an excellent dice roll. Which was what the primary goal here was.

What? No. Or not really. The whole point of using 3d6 vs d20 is that it's incredibly hard to get an "excellent dice roll" with 3d6.
I haven't checked this guy's math, because that's the whole point of googling it rather than doing it myself, but it seems right. http://www.thedarkfortress.co.uk/tech_reports/3_dice_rolls.htm#.VPyWe_n4WRI


Last session, I had a first level Rogue attempt to steal from our party's 7th level Ranger. A Spot skill of 18, versus the Sleight of Hand skill of 4. And yet, despite the incredibly low odds, he had his coin purse stolen. Because he rolled really crappy, and I rolled really well. Now, yes. As the DM, I can fudge that. But I don't want to. The mechanics allow for that to happen. And, after listening to him rant about losing 10,000 gp because of luck, I have to agree. That IS stupid. That's really stupid. Hence trying to improve the mechanic. One that makes opposed skill checks primarily dependent on the character, and not luck.

I'm not giving up. There's some way to make them more fair. I just haven't found it yet.

Just swapping out d20 for 3d6, let's see what you have here. Your Ranger has a Spot of 18, the Rogue has a Sleight of Hand of 4. The Rogue needs a 14 or better on 3d6, which he'll get less than 15% of the time. (Compared to 35% of the time on a d20)

And by the way, it's not "just luck." It's carelessness. Why is an experienced adventurer carrying his good loot in an easily-accessible coin purse? You keep that stuff in an inside pocket somewhere in your masterwork adventurer's outfit, and have an obvious decoy coin purse with 6 cp and a folded paper with Explosive Runes. (Or, if you're my good-aligned cleric, a patronizing note telling you to make better choices wrapped around a piece of candy).

johnbragg
2015-03-08, 01:50 PM
Wait a minute, I think the problem is that we're both reading the skill checks wrong here. Or at least I did.

The Rogue is rolling to swipe the purse vs a DC 20. With 4 ranks and no modifiers, he needs a 16 or better--he has a 25% chance. Let's say he rolled a 20, so the Ranger has to beat a 24 to make his Spot check and catch the thief. With a +18 in Spot, Ranger needs a 6 or better, which he has a 75% chance of.

Sometimes luck is part of the game. (And sometimes "bad luck" is what happens when lack of preparation meets adversity, in the form of an NPC's nat 20).

Oh, modeling all of this with 3d6 instead of d20.

The Rogue is rolling vs DC 20, needs a 16 or better--4.62% chance.

Let's say it's a different Rogue 1 with some modifiers (Dex, race, feats, whatever) and the Ranger has to make a 24 spot check. Ranger needs a 6 or better, which he has a 95% chance of.

Seerow
2015-03-08, 01:58 PM
Like Grod said, a dice pool doesn't work too well with as huge a modifier range as you have for 3e. Because having a potential dice pool of like 40+ just isn't going to work well.

For 5e I think a dice pool would work. Instead of d20+x, roll xd6 count successes. That makes each point much more valuable, lets a character starting off with 3d6 or so competent and makes the high level experts with 11d6 or so really incredible by comparison. Turn the Rogue/Bard skill benefit into a lower target number for success rather than doubling skill bonuses, and you've got a pretty workable system. The problem is that people get grumpy when you try to take the d20 out of skill rolls.




For a while I was working on an alternate skill system for 3.5 where rather than 0-23 ranks, you have 0-8 ranks. Max rank at level 1 is 2, and rank cap increased every 3 levels thereafter (4, 7, 10, 13, 16, 19). Each rank had two effects:

1) Translates into a +3 bonus on skill checks related to that skill.
2) Skill tasks were given ranks in addition to DCs. For each rank you are higher than the task, roll an extra d20, take best. Each lower roll an extra 20 and take lowest. Max of ~3 ranks of difference (ie rolling 4d20+x take best/worst) before auto success/auto failure kick in.

The idea was to make skills come in fewer more discrete chunks, with most characters gaining ~2 skillpoints per level. But I ran into trouble coming up with suitably high level abilities for the upper end of the skill DCs and eventually dropped it.

Djinn_in_Tonic
2015-03-08, 04:29 PM
For 5e I think a dice pool would work. Instead of d20+x, roll xd6 count successes. That makes each point much more valuable, lets a character starting off with 3d6 or so competent and makes the high level experts with 11d6 or so really incredible by comparison. Turn the Rogue/Bard skill benefit into a lower target number for success rather than doubling skill bonuses, and you've got a pretty workable system. The problem is that people get grumpy when you try to take the d20 out of skill rolls.

Another option would be to create something like Skill Mastery tiers: Advanced, Mastered, or Legendary, for example.

Advanced: Roll 2d20 instead of 1d20 when making a skill check, and choose the best result.
Mastered: Roll 3d20 instead of 1d20 when making a skill check, and choose the best result.
Legendary: Roll 4d20 instead of 1d20 when making a skill check, and choose the best result.

These could be granted as part of the leveling process (pick 1 skill to advance one stage every other level, for example), by class features, by feats, or some other way entirely.

You'll end up with the same sort of high-low ending values that the normal system would get (at least in 5e), but you'll find that highly trained characters will much more reliably be able to tackle tasks of moderate to high difficulties.

An added bonus to this is that players get both the benefit of getting to pick the best roll (which feels really good), and the benefit of it not being to far removed from the d20 system they're already familiar with.

The downside is the same as with any of these proposals: skilled characters will much more frequently win when opposed by characters of lesser skill, which may or may not be desirable depending on your specific group's playstyle.

Note: While this helps resolve the issue at mid-high levels, it does still leave low levels a bit dependent on the random elements of a d20.

johnbragg
2015-03-08, 06:07 PM
An added bonus to this is that players get both the benefit of getting to pick the best roll (which feels really good), and the benefit of it not being to far removed from the d20 system they're already familiar with.


Especially if they're playing 5th, and this sounds almost exactly like the Advantage mechanic.

Djinn_in_Tonic
2015-03-08, 08:47 PM
Especially if they're playing 5th, and this sounds almost exactly like the Advantage mechanic.

Completely intentional. :D

Zireael
2015-03-09, 02:20 AM
I like Djinn in Tonic's idea, somewhat like the luck/rerolling feats (roll again and pick best :D)