Anyone tried the 3d6 variant?

[Altair_the_Vexed]

There's a 3d6 in place of d20 variant rule in Unearthed Arcana (http://www.d20srd.org/srd/variant/adventuring/bellCurveRolls.htm). I've always found it a attractive - making the bonuses count for more, and randomising less - but I'm wary of it... Can't really explain why, it just feels a bit wrong.

Anyway, what I was wondering was: has anyone here in the Playground been in a game using that variant rule, and how did it play out?

(I'm looking for opinion and gut feelings - I know the maths fairly well.)

[AttilaTheGeek]

I've never used it, and I don't plan on doing so. Replacing 1d20 with 3d6 makes rolls at the end of the spectrum much less likely (as you know), making combat much more predictable. Of course, if that's what you're going for, it'll work well for you.

[koboldish]

I haven't tried it, but it doesn't feel right for me either. It's the d20 system, after all.

[silverwolfer]

It is rounding rolls , using a bell curve method.

[Zahhak]

I know the maths fairly well.

Really? Because the maths behind a 3d6 system is basically a bunch of crap.

With a 3d6 system you'd reduce the range from 1 to 20 on the 1d20 to 3 to 18, all while changing the average roll from 10.5 on 1d20 to... 10.5. So with a 1d20 your average is half way through the range of possibilities, while with a 3d6 system it's 2/3s the way through the range. So if you're using a system where you want to get high numbers (in DnD you want a 20 for a critical), you're basically golden, but if you are using a system where you want low numbers (HERO system), you're basically screwed.

I don't know how the 3d6 optional rule works in DnD, but in HERO it's basically the most basic and the most broken rule.

[Kelb_Panthera]

Really? Because the maths behind a 3d6 system is basically a bunch of crap.

With a 3d6 system you'd reduce the range from 1 to 20 on the 1d20 to 3 to 18, all while changing the average roll from 10.5 on 1d20 to... 10.5. So with a 1d20 your average is half way through the range of possibilities, while with a 3d6 system it's 2/3s the way through the range. So if you're using a system where you want to get high numbers (in DnD you want a 20 for a critical), you're basically golden, but if you are using a system where you want low numbers (HERO system), you're basically screwed.

I don't know how the 3d6 optional rule works in DnD, but in HERO it's basically the most basic and the most broken rule.

ummm.......... No.

10.5 is the same average and its equidistant from its extremes; exactly 7.5 more than 3 and 7.5 less than 18.

All it does is changes the statistical curve to favor the average over the extremes. The reduction in random chance ultimately favors the players in the long run though it -may- be a severe detriment in any single encounter, depending on the relative difference between the modifier and the target number..

[Xerxus]

I really don't know what the above poster means. 1-20 and 3-18 have the same average and the average is right at the median in both. The only differences are a slightly different range of possibilities and a bell curve vs an even spread.

[Zahhak]

10.5 is the same average and its equidistant from its extremes; exactly 7.5 more than 3 and 7.5 less than 18.

HOLY CRAP! How did I screw that up!?

Whatever, never mind.

[Blisstake]

So, yeah, the math works...

But it's math not intended for the d20 system, which kinda screws things up with criticals and critial misses and so forth. Also, the d20 system is supposed to be fairly dynamic, with a very wide range of rolls, each with an equal chance. With 3d6, you're more likely to roll near the center which makes bonuses count for a lot more, even more than they already do (which is quite a lot).

So I like 3d6 overall, but it's not really good for d&d.

[Lonely Tylenol]

I love D6 systems--I have a particular fondness for Green Ronin's Dragon Age RPG (although it has some flaws itself), and am building a system myself, bit by bit, which incorporates, among other things, the 3d6 system.

As far as D&D goes?

The system is built from the ground up on the d20 model, and a lot of things would probably need to be tweaked in order to make 3d6 fit comfortably--for example, critical hits and misses no longer function as they should based on the d20 roll, as they're both on the extreme end of the curve and involve numbers that don't exist on a 3d6 roll, and "take 10" and "take 20" rules would probably need to be modified somewhat--but it can be done, and it gives a group uncomfortable with random swinginess some comfort in the curve, so I'd be all game for it.

[Xerxus]

So, yeah, the math works...

But it's math not intended for the d20 system, which kinda screws things up with criticals and critial misses and so forth. Also, the d20 system is supposed to be fairly dynamic, with a very wide range of rolls, each with an equal chance. With 3d6, you're more likely to roll near the center which makes bonuses count for a lot more, even more than they already do (which is quite a lot).

So I like 3d6 overall, but it's not really good for d&d.

They do provide acceptable critical ranges.

[Kelb_Panthera]

So, yeah, the math works...

But it's math not intended for the d20 system, which kinda screws things up with criticals and critial misses and so forth. Also, the d20 system is supposed to be fairly dynamic, with a very wide range of rolls, each with an equal chance. With 3d6, you're more likely to roll near the center which makes bonuses count for a lot more, even more than they already do (which is quite a lot).

So I like 3d6 overall, but it's not really good for d&d.

The variant has listed fixes for the critical threat ranges and critical misses are (bad) houserules anyway.

As for whether it's good for the game or not; that depends entirely on what sort of feel you're going for. Because extreme strokes of luck are marginalized but luck is not eliminated as a factor those rarer crits will feel more spectacular for being so much rarer, for instance.

[Lonely Tylenol]

The variant has listed fixes for the critical threat ranges and critical misses are (bad) houserules anyway.

That depends on how you define a critical miss: that somebody automatically fails on a natural 1 [for attack rolls and saves] is neither a houserule nor bad, but it's something I'd define as a "critical miss".

[Pesimismrocks]

While the 3D6 system can work mathematically, and it makes it more 'realistic' in that there is more focus on things like ranks, it simply takes away the dynamic.

In my experience the random chance of a cleric critting 3 times in an arm-wrestling contest with a barbarian, a fighter being out drunk by a wizard or even a ninja failing to dodge a fireball where a slaad does; it's just a lot more enjoyable than a fairer system based on averages.

[Altair_the_Vexed]

Thanks for all the opinions folks - but had anyone actually USED the rule or not?

[Pesimismrocks]

Yes. Me and my regular group tried it out several months ago. Nothing could hurt the cleric and we barely hurt tougher monsters. Combat quickly became drawn out and dull

[DaTedinator]

I have, and I actually posted about it a few years ago. It was a pretty lengthy post, though, so rather than copy/paste it, I'll just give you a link (http://www.giantitp.com/forums/showthread.php?t=159247).

Now that a few years have passed and we played a good bit more in that campaign, as well as another campaign where we used 2d10 instead of 1d20, I have some additional thoughts on the matter.

3d6 in combat isn't great, and I don't think I'd do it again, even in a social-focused game. It's just not random enough; combat isn't very chaotic, and I like my combats chaotic. If you want a bell curve in combat, you can do 2d10, which is better. I think I personally still prefer 1d20, but there are some in my group I know prefer 2d10. It keeps things tending towards 11, but still with a good deal of randomness.

Outside of combat, though, I love 3d6. Like, for skills and such? It's great, for all the reasons I listed before.

For an upcoming campaign, I think what I'm going to do is use 1d20 and 3d6, depending on the situation. If things are calm enough where you could take 10 on things, you roll 3d6 (you could roll 1d20 if you wanted to). If you can't take 10, you roll 1d20.

That way, I'll get the randomness and chaos of combat, and also the randomness in skill checks because my players won't just be taking 10 all the time.

I hope that's helpful.

[Ashtagon]

http://cdn.memegenerator.net/instances/400x/34965525.jpg

[Clericzilla]

http://cdn.memegenerator.net/instances/400x/34966856.jpg


EDIT: Fixed!

[sreservoir]

the 3d6 variant specifically, no, but I've done 3d20 mid (http://anydice.com/program/6d0), both with and sans adjustments to threat ranges.

in the former case, map 15-20*→ 14-20, 17-20 → 15-20, 18-20 → 16-20, 19-20 → 17-20, 20 → 18-20.

but 3d20 mid is parabolic, not multinomial; much less steep a curve; I imagine a multinomial distribution (and especially 3d6, because 3d6 cuts off) necessitates adjusting those ranges.

anyway, it slows down play somewhat, but that's expected; in general, it seems to reduce tension, and it makes those encounters with fewer numerically more powerful enemies disproportionately more dangerous and encounters with more less powerful enemies less dangerous; can't get away with as large a gap between pc and npc capability.

on the whole, though, I'm quite pleased with it.

it ... seems like it would be more satisfying to migrate to a curve with something that actually uses d20 as the only resolution mechanic (I, uh, haven't managed to run or play m&m or true20 -- I mean, I pulled the 3d20 mid mechanic out of an m&m book, but).


3d6 in combat isn't great, and I don't think I'd do it again, even in a social-focused game. It's just not random enough; combat isn't very chaotic, and I like my combats chaotic. If you want a bell curve in combat, you can do 2d10, which is better. I think I personally still prefer 1d20, but there are some in my group I know prefer 2d10. It keeps things tending towards 11, but still with a good deal of randomness.

can I complain enough about calling 2d10 a "bell curve"? (http://anydice.com/program/392)

multinomial distributions with n = 2 aren't curves >((( they're just piecewise linear.

(also has a centre +0.5 from d20, which biases in favour of whoever is rolling attack rolls, saving throws, &c., so)

[javijuji]

I have used it and although it is better it takes away the d20 magic.

[DaTedinator]

can I complain enough about calling 2d10 a "bell curve"? (http://anydice.com/program/392)

multinomial distributions with n = 2 aren't curves >((( they're just piecewise linear.

(also has a centre +0.5 from d20, which biases in favour of whoever is rolling attack rolls, saving throws, &c., so)

Haha, sorry, I don't know very much about... statistics? Hah. I don't even know what form of math the term "bell curve" comes from. :smalltongue: And I definitely didn't know what it specifically meant. Regardless, I believe my meaning was clear.

As for the +0.5, rest assured, it really doesn't make a noticeable difference in play. And even if it did, it favors the party.

[SowZ]

It works better alongside the facing variant, (makes it easier to get some bonus against high ac enemies/enemies against you when you have high ac,) and the armor as dr variant, (keep ac low.) Not sure if it would work better at low or mid levels.

[Draz74]

I don't even know what form of math the term "bell curve" comes from. :smalltongue: And I definitely didn't know what it specifically meant.

It refers to the way the probability distribution looks like an old-fashioned bell (http://anydice.com/program/1), like the Liberty Bell.

[Mighty_Chicken]

Ok, let's take first levels play for an example.

With d20, the barbarian with +7 attack roll has 70% chance to hit that ac 14 minion. Powerful, but that's also 30% miss chance. Because a 2 is just as possible as a 15. So if he's fighting 4 of 14 AC minions, he's probably gonna miss at least once and that could mean his death. We intuitivelly know 4 14 AC enemies might be too much for that 1st level barbarian.

Then again: the barb could only screw up with 6 results: 1, 2, 3, 4, 5, or 6.

What with 3d6? Can he only screw up with 4 results (3, 4, 5, 6)? Nooope. He can screw up with 19 results. 1-1-1, 2-1-1, 1-2-1... and so on. Count it. It's 1 result for 3, 3 for 4, 6 for 5 and 9 results for 6. He can screw up in 19 ways - but he can success in 197 ways. Because with 3d6 you don't have 20 possible results, you have 216 of them, and the middle ones are the most common. This means now our barb has only an 11% chance of screwing up (19/216).

In other words, each 14 AC henchmen just got 3 times less dangerous. Would you be afraid of not hitting 4 such minions in 4 turns? I wouldn't.

That's what 3d6 does: any superiority means miles. So forget everything you know about balance and CRs. A boss could kill your whole party by himself, just after they murdered 300 of his henchmen painlessly.

There a second problem: any character with big bonuses are much more powerful. "Bah, who cares about that +3 bonus he has to appraise?" Nope, in 3d6, he's the god of appraisal. Barbarians are specially ridiculous here. But any +2 or -2 - damn, even +1 or -1 - buffs or nerfs will change the outcomes of a battle completely. With d20, any +1 means a +5% buff. In 3d6, it's much harder to calculate. A +1 to tough test might help almost nothing, and a +1 to an easy test is just making things more auto win. +1 to moderate tests are a BIG difference of over 10%.

3d6 has 216 possible results. And low results are as rare as natural blond curly hair. So people with high bonuses are invencible. And high results are just as rare - people with low bonuses are useless.

2d10, however, has 99 results (and a 01 ain't one of them). The possibility for the smallest and highest result (02 and 20) are 1% rather that 0,45% (no kidding, that's how 3d6 is).

D&D is madness with 3d6. With 2d10, it isn't balanced either, but it is still D&D.

I'm using 2d10 for skills in my campaign. Players love it,because they know if they suck or if they excel at doing things. I won't use it for battles, though, because I'd need to rebalance the WHOLE game from ground zero. From BaB to feats and monsters. But it's yout call if you wanna try.

[Captainspork]

I haven't tried this out in a complete campaign, but I've done some of the number crunching and played out a few as senarios to see how it went. The numbers stuff has been explained, so I will say that if you've tried it at all on your own(and you should if you haven't), you know that "+1,2,etc" buffs feel boosted a lot, and it's very hard to scale encounters. As people have said, the system isn't designed for it. While fumbles can suck, I personaly think they are part of the "magic" of the game which can be fun if you play it right.

If you're committed to using it, I'd try a hybrid system, using 3d6 for skill checks and 1d20 for combat. I don't think I'll ever use 3d6 on its own though. Watching a player with 20 ranks in hide roll a 1 on a skill check can be pretty lame, but making a player or, worse, a monster, invincible in combat raises a much greater issue IMO.

[Mighty_Chicken]

You just made me notive something, Captainspork.

Chance of 1 in d20: 5%

Chance of 2 in 2d10: 1%

Chance of 3 in 3d6: 0,46%

So, do you want any natural result to be an automatic fail? With 3d6 or 2d10, you can tweak the chance of an automatic fail (or success) with some freedom

In 2d10: you can take that just 2 is an automatic fail (chance=1%) or that both 2 and 3 are (chance=3%, still less than d20)

In 3d6: you can take just 3 (0,46%), or both 3 and 4 (2%), or 3, 4 and 5 (4,7%).

Of course, you could also make automatic fails or successes more common. But I think this is interesting.