Anyone with experience rolling 3d6 instead of 1d20?

[Mystral]

Hey, y'all.

I have recently switched game systems with my group, from someone else running a game of TDE to Pathfinder. After a few sessions, the players expressed high dissatisfaction with the game system, namely, the importance and high randomisation of the d20. They felt that their bonuses weren't important enough in contrast to the roll of the die.

There isn't really much sense discussing someone's dislikes, so we switched to the bell curve rolls detailed here: http://www.dandwiki.com/wiki/UA:Bell_Curve_Rolls

The players were more satisfied with this, so we'll keep using it. The question is, what does this mean for me as a DM? Are there any important things I should know, any pitfalls? I don't want to wipe my party just because I fail to understand this different way of playing the game. if someone could enlighten me, I would be very gratefull.

[Angelmaker]

Less chance of criticals. Might need a few adaptations. Played with three d six in dragonage. Works. Liked it better too.

[Grod_The_Giant]

You're less likely to get lucky breaks-- either good luck or bad. If their bonuses don't suggest that the players can hit that skill check/save/AC, they probably won't. At the same time, they'll hit level-appropriate targets much more often. Weaker monsters will become slightly less so, while stronger will become slightly more so. The Unearthed Arcana (http://www.d20srd.org/srd/variant/adventuring/bellCurveRolls.htm) rules suggest fiddling with threat ranges, as well as changing rerolls into "4d6, drop the lowest."

[Mystral]

Threat ranges are already adjusted in those rules (a threat on 20 in d20 is a 16-18 in 3d6)

Do you mean that weaker enemies will become stronger?

[Sacrieur]

I'd use 3d8. 3 is a 1, 23-24 is a 20. Everything else is 1:1.

[Mystral]

I'd use 3d8. 3 is a 1, 23-24 is a 20. Everything else is 1:1.

Not an option, that would place an even greater emphasis on the die roll than with 1d20.

[Chronos]

Do you mean that weaker enemies will become stronger?
No. Consistency benefits the stronger combatant (which is usually the PCs). Let's say we have a PC fighting a significantly weaker monster. The PC will probably hit... but there's a small chance she might miss. The monster will probably miss... but there's a small chance he might hit. More consistent rolls mean that both of those small chances get even smaller.

[BowStreetRunner]

I've played in a 3.5 game where the d20 was replaced with 4d6-4 (result of 0-20 with 0=fumble) before. My take on it was that it falls somewhere between the d20 system and Amber diceless RPG.

In Amber, all that matters are your stats - if they are higher you always win, if lower you always lose. There is no random chance. Outcomes are always predictable, so no one ever really takes any chances. Instead, they always play to their strengths and find ways to accomplish their goals with the abilities in which they have the advantage.

With Bell Curve Rolls unexpected results can still occur, just not as often. The larger the number of dice used, the less random and more regular your results (4d6-4 is less random than 3d6). Both players and DMs need to curb their expectations a bit and take fewer chances. It remains a good idea to play to your strengths and find situations that provide an advantage. Yet sometimes you can still just get lucky - for good or ill.

[Sacrieur]

Not an option, that would place an even greater emphasis on the die roll than with 1d20.

What do you mean?

The distribution curve is still there.

[Snowbluff]

I have. There's no point. Bell curves are worse for newer plays, blah blah blah.

[Mystral]

What do you mean?

The distribution curve is still there.

But the rolls would now be higher, meaning that the rolls would still outstrip their bonuses in importance.

Also, it would wreak havoc with game balance if the mode of the die roll would be 13,5 instead of 10,5.


I have. There's no point. Bell curves are worse for newer plays, blah blah blah.

Well, the point is that my players have more fun using them.

[Elkad]

The main issue with multiple dice is that tiny changes have a huge impact on the odds. And the more you add, the worse it gets.

If you need an 11 to hit, and your target gets a +2 AC bonus (so you need a 13), your chance to hit drops from 50% to 25%.
If he got a +4 AC bonus instead, you'd drop to 9%

+5 to hit is the same as "don't roll a 1" in practice, and +5 to AC becomes "everyone needs a 20"

[Chronos]

Quoth BowStreetRunner:

With Bell Curve Rolls unexpected results can still occur, just not as often. The larger the number of dice used, the less random and more regular your results (4d6-4 is less random than 3d6).
Not true. More dice of the same type is always more random (by which I assume you mean greater standard deviation). The standard deviation grows more slowly than the mean, but then you pull the mean back down with that nonrandom -4. 3d6 is less random than 1d20, but that's because the dice you're using are smaller, not because there are more of them.

The mean of an n-sided die is (n+1)/2. Its variance is (n^2 - 1)/12. When you roll multiple dice and add them, the means add. When you roll multiple dice and add or subtract them, the variances add. The standard deviation is the square root of the variance.

Thus, for instance, a d6 has a mean of 3.5 and a variance of 35/12. 3d6 has a mean of 10.5 and a variance of 35/4, for a standard deviation of 2.958. 4d6-4 has a mean of 10 and a variance of 35/3, for a standard deviation of 3.416.

[danzibr]

Huh. I'm surprised nobody mentioned 2d10. It's how my group plays. Much prefer it to 1d20.

[Snowbluff]

Well, the point is that my players have more fun using them.

From what you said, I get a pretty telling image of your players:

1) They don't respect the proper RNG of a twenty sided die.

2) They aren't very good optimizers.

3) Ergo, the d20 system (3rd, 3.5, PF) might not be the right system for them if the above are true.

[OldTrees1]

I haven't tried 3d6 but I did try the opposite (1d6+2d8-2, -10 if over 10) as a gimmick for one of my characters. It was loads of fun for a while(8 sessions) but I noticed I started to miss the grass on the other side.

1d6+2d8-2 (1-20 bell curve)probability
1: 0.26
2: 0.78
3: 1.56
4: 2.60
5: 3.91
6: 5.47
7: 7.03
8: 8.59
9: 9.64
10: 10.16
11: 10.16
12: 9.64
13: 8.59
14: 7.03
15: 5.47
16: 3.91
17: 2.60
18: 1.56
19: 0.78
20: 0.26

[Barstro]

Using multiple dice transforms the straight line of a d20 (each +1 is a +5% chance) to a bell curve. The rolls are much more likely to hover around the middle than be an outlier. The rules already seem to take this into account for the Critical Hit aspect of it, but that's only part of the picture.

2d10 is an easy way to look at it.
There is one way to get a 2 or a 20
Two ways to get a 3 or 19
Ten ways to get an 11

3d6 makes the bell curve even steeper.

This means it will be much harder to hit someone if you require a 12+ as a result by rolling 2d10 than 1d20. (The opposite is also true). It makes hard fights even harder.

It's not a bad system, but I think other things need to be factored in.

[Mystral]

From what you said, I get a pretty telling image of your players:

1) They don't respect the proper RNG of a twenty sided die.

2) They aren't very good optimizers.

3) Ergo, the d20 system (3rd, 3.5, PF) might not be the right system for them if the above are true.

Thank you for your input, none more is required.

[Mystral]

Using multiple dice transforms the straight line of a d20 (each +1 is a +5% chance) to a bell curve. The rolls are much more likely to hover around the middle than be an outlier. The rules already seem to take this into account for the Critical Hit aspect of it, but that's only part of the picture.

2d10 is an easy way to look at it.
There is one way to get a 2 or a 20
Two ways to get a 3 or 19
Ten ways to get an 11

3d6 makes the bell curve even steeper.

This means it will be much harder to hit someone if you require a 12+ as a result by rolling 2d10 than 1d20. (The opposite is also true). It makes hard fights even harder.

It's not a bad system, but I think other things need to be factored in.

I like this 2d10 thing. I think I will offer it to my players. 1d6+2d8-2 seems too complicated.

[MyrPsychologist]

I prefer 1d20.

I understand the feeling that there is more luck and that it can be unfun to have a lucky/unlucky roll of the dice just completely screw up some plans. But I feel like that is part of the fun. Without that air of randomness and unpredictability things become much more boring and players probably won't feel incentivized to take risks because there just isn't a point.

I feel like the normal system, tempered with a solid GM that doesn't throw stupid fumbles at the party is the best way to go. Especially when they use the mindset that even failed rolls should advance the plot in some way. After all, a story would be exceptionally boring if the heroes never failed and met with incredible success everywhere they go.

[Mystral]

I think I prefer 1d20, too. But those players are new to d20 and used to something different.

TDE (The Dark Eye) uses a rather more complicated system for rolling on skills, using 3d20 to try and roll under three attributes contributing to the skill, and skill points to even out the numbers rolled over those. It's just what they are used to, but converting the module we are playing to DSA is out of the question, as is me playing a TDE module, which I loathe with a vengeance.

[Grod_The_Giant]

From what you said, I get a pretty telling image of your players:
Rude, harsh, and unfair. The d20 is a very swingy RNG method. Fact. For most of the game, before your modifiers start vastly outstripping the dice, you will fail checks you should have succeeded at. You'll go entire nights without rolling about a 5 (or at least feeling that way). It's not wrong to like that, but it's also not wrong to feel frustrated by it. There's nothing wrong with wanting a more consistent roll. In fact, I'd think it would encourage risk taking, because it means you can actually rely on your skills.

Elkad has a good point about the relative value of bonuses, though-- you might want to cut all numerical modifiers in half or something. Or go to a 2d10; that feels pretty random while still trending towards average values.

[Vrakk]

I've never played with anything other than d20 in D&D. When playing other systems with d6's or d10's I liked it but it seems weird for this game. In big situations the entire group will slightly raise out of their chairs to see the roll of the d20 - I think it's this feeling than makes the game so fun. While adding more dice removes the chance of failure I think it also takes away a huge part of the anticipation. And as a horribly lucked die roller - I almost always find a way to take the Luck Domain so I don't die, ha.

[Mendicant]

The main issue with multiple dice is that tiny changes have a huge impact on the odds. And the more you add, the worse it gets.

Yeah, this is the biggest thing to keep an eye on. I'd watch out with encounter building--the game assumes you can pit the party against enemies with higher CRs than a non-d20 RNG might allow.

Are you familiar with anydice (http://anydice.com/)? I've found playing with it is a really good way to get a more intuitive, visual sense of what the consequences of various die manipulations are going to be.

[Snowbluff]

Rude, harsh, and unfair. The d20 is a very swingy RNG method. Fact. For most of the game, before your modifiers start vastly outstripping the dice, you will fail checks you should have succeeded at. You'll go entire nights without rolling about a 5 (or at least feeling that way). It's not wrong to like that, but it's also not wrong to feel frustrated by it. There's nothing wrong with wanting a more consistent roll. In fact, I'd think it would encourage risk taking, because it means you can actually rely on your skills.


I'd say so, but it's PF. Getting +9 to a skill in your area of expertise (or sometimes out of it, with Traits like Student of Philosophy replacing the stats) is easy at level 1 (+3 Stat, +1 stat bonus from race, +1 trait, +4 from putting in a skill rank, and this is before racial and class skill bonuses). 10.5 is the average roll on a d20, so you start off almost meeting the dice's contribution, which is why you had to say "vastly outstrip."

Now, in my experience, players have an excessive amount of bravado by default. If they think they'll succeed at everything, and then go rolling for everything, getting themselves into trouble and causing messes for everyone else. This okay for a silly game, but one with some gravitas or an "epic" quality, it can really kill the tension if the characters have the upper hand all of the time. :smalltongue:

Not to mention if they're are actually in a situation where it's a risk, and 11 won't make the check, anyway.

[AmberVael]

I have some experience with this kind of system, particularly, I have some experience using it in some circumstances which weren't quite as suited to it- and let me tell you, if the system isn't designed around it from the start, or if even just a little bit is off, then it can go bad really quickly.

D&D isn't designed around bell curve probability, and characters can have wildly divergent stats and bonuses, so to make 3d6 work you'd frankly need to overhaul the system. Elklad points out exactly where problems start coming in with D&D. Think of all the situational bonuses, and then stuff like one character having a low save and another character having a high save. Throw something at them that stuns, and suddenly one character gets no turns ever, and as far as the other character is concerned stunning doesn't even exist. Basically, throwing an encounter at your players that challenges all of them becomes really freaking hard, because what is easy to one of them may be certain death to another.

A good example of how it should work is Fate. Fate uses four fudge dice, and tends to have very small modifiers/stats, because it is built around the idea that even one point of difference will matter a lot to a character.

[Ssalarn]

***

D&D isn't designed around bell curve probability, and characters can have wildly divergent stats and bonuses, so to make 3d6 work you'd frankly need to overhaul the system. Elklad points out exactly where problems start coming in with D&D. Think of all the situational bonuses, and then stuff like one character having a low save and another character having a high save. Throw something at them that stuns, and suddenly one character gets no turns ever, and as far as the other character is concerned stunning doesn't even exist. Basically, throwing an encounter at your players that challenges all of them becomes really freaking hard, because what is easy to one of them may be certain death to another.

***

This is what I was going to point out as well. The entire D&D/Pathfinder system is designed around that 20 point variance, and changing it will have numerous consequences, not the least of which is creating a situation where low save characters will almost always fail their save, and high save characters will almost always succeed. This has really unfortunate effects, like further nerfing Fighters and Rogues, whose lives were hard enough already. The 3d6 system will seem more beneficial at levels 1-3ish, where the party's biggest dissatisfaction is going to come from missing their attack rolls more frequently on a d20, but as the game progresses and saving throws become a more frequently used defense and static bonuses pile up in strong areas on each character it's going to become much less fun as it transitions to "characters/creatures who are strong against certain effects will always win, characters who are weak will always lose". So amongst other things, the 3d6 method also vastly exacerbates martial/caster disparity, since casters are the ones typically capable of choosing which defense they target.

[Flickerdart]

The problem with the d20 isn't that it's swingy, it's that you don't roll enough of them. A single roll spells the difference between success and failure in many enterprises - skill checks, saves vs death, etc.

[lsfreak]

In order to really switch to 3d6, you as the DM need to be willing to very carefully balance monsters as well as characters. People with high or low modifiers will regularly be in big trouble, or face no danger. Oppositely, when you're used to saying the PCs will need between 12 and 16 to hit the monster, a relatively good spread a bit on the tough side, that's now much, much more difficult for the person who needs the 16 while being no more difficult for the one who needs the 12. You basically have to on-the-fly rebalance the game, which is why I gave up on it. 2d10 might be a compromise that doesn't upset the game as much, and I've considered using 3d6 for only skill checks and d20 for everything else, keeping a lot of the swinginess in combat (and making it easier to balance), while making it so that skill ranks are more important and someone with a +10 doesn't still fail a D15 check a quarter of the time.

[Zakerst]

So I have some experience playing in a system based around 3d6 and its fine, I also have experience playing in d20 systems and d100 systems.
Now as for playing DND as a 3d6 system you're going to have some problems where normally a '10' is the tipping point in d20 in 3d6 that 10 + or - 1 or 2 is going to eat up most of your roles, meaning an encounter based in the d20 system that's looking for a '15' to hit it (i.e. a 1/4 chance) is going to be way less in a 3d6 system. Not to mention there're going to be many times where a lucky role to save yourself or even just above average which could be semi-reasonable on d20 just aren't going to come up in 3d6.

likewise Finding a decent challenge is going to be hard and you'll likely have to tweak each encounter and mob a bit to match the expected results. This is also going to mean some challenges that would be difficult are going to be neigh impossible, (don't use anything form MM3) and in general the CR system is going to be even more borked.

A third out come is that flat bonuses are going to be really attractive, which might not be horrible depending on your group, but gods help you if you should ever fall behind WBL guidelines.

In short it makes the game a lot less swingy but also further polarizes what is doable and what is not.

For a quick explanation lets look as some math:

On a d20 each number has a flat 5% chance of showing up (1/20)

On 3d6 you get 3s (equivalent of 1s I'll assume for now about .46% of the time), 4s 1.39%, 5s 2.78%, 6s 4.63%, 7s 6.94%, 8s 9.72%, 9s 11.57%, 10s 12.50%, 11s 12.50%, 12s 11.57%, 13s 9.72%, 14s 6.94%, 15s 4.63%, 16s 2.78%, 17s 1.39%, 18 0.46%

now to go back to my previous example of 10+-2 in d20 that's going to account for 25% of your roles, in 3d6 its about 57.89% even at 10+-1 in d20 its only 15%, in 3d6 its still 36.57%.

[Barstro]

In the end, all the rolls do the same thing.

If you have a situation where player should hit the NPC 40% of the time, it's 1) 12 or higher on a d20, 2) 13 (or so; I did not calculate it) or higher on 2d10, 3) 60 or higher on percentage dice, 4) 12 or higher on 3d6. No matter what you do, the DM needs to taylor each scenario to the occasion. The differences happen when bonuses vine into play. Bonuses mean a lot more in a bell curve.

Whatever system the players want is fine, so long as other changes are made accordingly. I still don't want to go against a BBEG with a bell curve array.

[BowStreetRunner]

With 3d6 here is the percentage of rolls that will result in each possible outcome. With 1d20 all outcomes are equal to 5%.

3 0.46%
4 1.39%
5 2.78%
6 4.63%
7 6.94%
8 9.72%
9 11.57%
10 12.50%
11 12.50%
12 11.57%
13 9.72%
14 6.94%
15 4.63%
16 2.78%
17 1.39%
18 0.46%

[Shackel]

Be sure to check if your group really wants a 3d6 system. Although Snowbluff put it a little too bluntly, looking for better optimization or simply starting at a higher level of power may be needed.

In other words, if you need the average roll to hit someone, let's say AC 20, it'll take an AC25 than that to make it a 25% chance to hit them.
In this it'll take AC23.

In a d20 system the difference between a high and low save is going from 50% to ~20% on an identical roll+other bonuses by level 20. In 3d6 it's going from ~50% to about 3% on an identical roll+other bonuses.

So be very, very careful about pitting them against creatures with higher-than-normal(even normal) DCs for saves or AC, skill checks that need a higher-than-normal DC, etc. Lucky breaks are almost a non-factor, now.

[danzibr]

Man, I've been meaning to post in this thread for a couple weeks, finally got around to it.

I want to add a bit to what's already been said. It's sort of been said, but not with in this manner.

Basically, using 2d10 or 3d6 instead of 1d20 is favorable (that is, works in favor of the players) for players with a decent level of optimization skillz. Check these examples. I'll be using 2d10 because the math is a bit easier.

1) Unoptimized character needs to roll at least a 16 to hit a bad guy. Using 1d20 he has a 25% chance of success. Using 2d10 he only has a 6% chance of success. Ouch.
2) Optimized character needs to roll at least a 5 to hit a bad guy. Using 1d20 he has a 75% chance of success. Using 2d10 he has a whopping 94% chance of success. Hurrah for him.

It's even more drastic with 3d6.

[HurinTheCursed]

I played with 3d6 for years in another system. It works very well to add more "realism" or more heroics. Characters are better at what they are supposed to be good and worse at what they are bad.

As said for d&d, changing the standard deviation from 5.8 to 2.9 makes a lot of difference in easy and difficult tests, not much in average difficulties. It means players who have characters that have a good answer to each problem are favored over those that are more static, ie the gap between a decently player magic user is broadened with a decently played mundane. It also favors players able to plan, to stack many small modifiers, hence good players and optimizers over casual players.

Characters will tend to be more confident in their abilities and can afford more rolls, more difficult rolls in their speciality since it's still easy. It favors specialists over jack of all trades since specialists can make more heroic stuff while semi specialized characters won't be any better and will fail these more heroic actions.
It also means the redundancy in the group can be a bit reduced (needed in d20 to make sure at least someone succeeds in the group) but not so much that a bad roll (that still happens) from the specialist means there is no back up solution. ie at least 2 characters should be good for each save to avoid some traps and save or suck.

A group that has no proper tool for a given problem or face a greater challenge will have to make semi difficult rolls which will make such encounters much more difficult.

For a similar encounter level, encounters with multiple low CR opponents will be easier than indicated by EL (unless they get the best of flanking), especially if they are all the same (because one good counter is more efficient in 3d6 than in d20), while very few high CR opponents should seem harder than the given EL. As said it also implies low EL are easier and high EL harder.

[P.F.]

Here are the probability distribution equivalencies.

3d6 = 1d20

3 = 1
4 = 1
5 = 1
6 = 1
7 = 2
8 = 3
9 = 5
10 = 8
11 = 11
12 = 14
13 = 16
14 = 18
15 = 19
16 = 20
17 = 20
18 = 20

Or, in reverse,

1d20 = 3d6

1 = 6 (or less)
2 = 7
3 = 8
4 = 8
5 = 9
6 = 9
7 = 10
8 = 10
9 = 10
10 = 11
11 = 11
12 = 11
13 = 12
14 = 12
15 = 12
16 = 13
17 = 13
18 = 14
19 = 15
20 = 16 (or more)

So the first thing people usually want to do with this information is calculate crit ranges. We can easily see that a d20 weapon with a 20 crit will threaten on a 16+ in 3d6; an 19-20 weapon will now threaten on a 15+, and the super-special 18-20 threat range weapons will now have a 14-18 crit range, depending on if you round the same way I do.

The more important implication, however, is that bonuses add up very differently in 3d6 than in d20. If I need a 15 or better to hit in d20, this is equivalent to only needing a 12 in 3d6, which with a +2 bonus I should on average, succeed. However, if I need a 15 to hit in 3d6, and the DC is increased by +2, I'm hosed: I have about a 1 in 50 chance of success.

I know that it's kind of annoying, especially at low levels, to have the swing be so much higher than the spread. But the game is balanced to have this spread. In the d20 system, bonuses at +2 and +4 are the norm, and one generally adds multiples of these bonuses to each roll. In games with multiple dice, like the 3d6 system, one will instead throw additional dice and/or have static +1 bonuses.

So in summary, I would suggest that crowbarring D&D/Pathfinder into 3d6 is probably inadvisable. It might be fun at lower levels, but your players might change their tune as their spreads outstrip the swing and they find that challenges meant to be in the 25%-75% success range suddenly turn into auto-pass for some characters and auto-fail for others.

If the random element is just too much for your players, I would suggest using the 'taking 10 to-hit" variant. This prevents constant misses and the wailing of "my skills don't matter" against weaker opponents, and against those foes who couldn't be hit with a 10+ anyway, you can point out to your players that they are LESS likely to succeed by rolling 3d6.

For those players which simply enjoy the mental math, I would recommend the Sacred Geometry (http://www.d20pfsrd.com/feats/general-feats/sacred-geometry) feat.

[Chronos]

Quoth Grod_the_Giant:

Elkad has a good point about the relative value of bonuses, though-- you might want to cut all numerical modifiers in half or something. Or go to a 2d10; that feels pretty random while still trending towards average values.
Except that the point of switching to 3d6 is that numerical modifiers don't do enough in the current system, and ought to do more. Consider three first-level characters: The strongest of half-orcs (Str 20), an average human (Str 10 or 11), and a runt of a halfling (Str 1). That seems like a huge range of strengths, right? But if you take a task of strength that the human can successfully do under stress about half of the time (DC 11), you'll find that that brute of a half-orc can do it only three quarters of the time, and the wimpy halfling can do it fully one quarter of the time. That doesn't seem right.

Change to 3d6 rolls, though, and the situation changes. The DC for the human to succeed half the time is still 11. But now, the half-orc succeeds 103/108 times, and the halfling succeeds only 5/108 times. That sounds a lot more consistent with the actual strength differences in play.

[atemu1234]

Huh. I'm surprised nobody mentioned 2d10. It's how my group plays. Much prefer it to 1d20.

Meh. You only have nineteen possible number results then. Though you have a hundred ways of receiving them.