PDA

View Full Version : 3d6 instead od d20



jinete
2010-10-09, 10:41 AM
Just had a call from my DM, he's thinking of implementing the 3d6 instead of d20 variant in our upcoming Red Hand of Doom campaign so he asked for my opinion on the matter and I thought I'd ask the boards for help.

I'm aware that it takes away some of the randomness from the game, e.g. crits will happen about 10 times less often. Has anybody actually played with this variant? Opinions?

We were also discussing using 2d10 instead of d20, does anybody have any experience with that?

Starbuck_II
2010-10-09, 10:54 AM
Just had a call from my DM, he's thinking of implementing the 3d6 instead of d20 variant in our upcoming Red Hand of Doom campaign so he asked for my opinion on the matter and I thought I'd ask the boards for help.

I'm aware that it takes away some of the randomness from the game, e.g. crits will happen about 10 times less often. Has anybody actually played with this variant? Opinions?

We were also discussing using 2d10 instead of d20, does anybody have any experience with that?

Wait, Crits happen more often. Remember Longswords are max -1, scimitars are -2.
So 18 = Crit with greataxe, 17 with longsword, and 16 with scimitars.

Easier to roll as 16 on a 3d6 than a 18 on a 1d20.

Eloel
2010-10-09, 10:59 AM
I've read a variant somewhere that let you use 4d6-4 and d20 interchangably (use whichever you wish for each roll). By calculations, 4d6-4 averages slightly lower, has a chance to roll a 0, but is more consistent around 10. Sounded fair to me.

Kylarra
2010-10-09, 11:05 AM
Wait, Crits happen more often. Remember Longswords are max -1, scimitars are -2.
So 18 = Crit with greataxe, 17 with longsword, and 16 with scimitars.

Easier to roll as 16 on a 3d6 than a 18 on a 1d20.Actually...

{table]Old Threat Range | New Threat
20 | 16-18
19-20 | 15-18
18-20 | 14-18
17-20 | 14-18
15-20 | 13-18[/table]
http://www.d20srd.org/srd/variant/adventuring/bellCurveRolls.htm

snoopy13a
2010-10-09, 11:07 AM
Wait, Crits happen more often. Remember Longswords are max -1, scimitars are -2.
So 18 = Crit with greataxe, 17 with longsword, and 16 with scimitars.

Easier to roll as 16 on a 3d6 than a 18 on a 1d20.

There's a 15% chance to get a 18+ on a d20.

There's about a 4.7% chance to get a 16+ on 3d6.

It is easier to get a 20 on a 1d20 than a 16 or better on 3d6.

http://sullivan.pgh.pa.us/~ksulliva/ralph/dnd-stats.html

Greenish
2010-10-09, 11:15 AM
I've been itching to try it out. It means more average rolls, less extremely good or bad ones, which in turn makes modifiers more important, especially at lower levels.

That site is very interesting, snoopy. According to it's numbers, the probability of critting with a keen rapier is 25.9% with 3d6, or 30% with 1d20.

So yeah, crits are less frequent.

Kaww
2010-10-09, 11:21 AM
I've read a variant somewhere that let you use 4d6-4 and d20 interchangably (use whichever you wish for each roll). By calculations, 4d6-4 averages slightly lower, has a chance to roll a 0, but is more consistent around 10. Sounded fair to me.

Slightly? Chance of rolling 20 is 1/6^4 (1/1296) in this variant, while it's 1/20 on d20. How is this a slight alteration? I once played with a DM that said we could roll d20 for our stats(1,2,3=3; 18,19,20=18). Those of us who knew our math and felt lucky tried their luck. One guy had three 18s... :smallbiggrin:

dobu
2010-10-09, 11:23 AM
in my group we do a '3d20 take middle roll' system, with slightly altered crit ranges. Works quite good so far. Much more predictable, while still random.
the probability isn't a bell function, it's like a hyperbolic function...

but as I said, quite a good system so far :-)

Eloel
2010-10-09, 11:34 AM
Slightly? Chance of rolling 20 is 1/6^4 (1/1296) in this variant, while it's 1/20 on d20. How is this a slight alteration?

Averages.

4d6 = 14 average.
4d6-4 = 10 average.
1d20 = 10.5 average.

0.5 is slight enough.

Kaww
2010-10-09, 11:42 AM
Chance of rolling 0: 1/1296
Chance of rolling 1: 4/1296
Chance of rolling 20: 1/1296

It's 5:1 for 1 or less, it should be even (translated to d20 1,2,3,4,5 - critical fail, 20 critical success)...

Eloel
2010-10-09, 11:51 AM
Chance of rolling 0: 1/1296
Chance of rolling 1: 4/1296
Chance of rolling 20: 1/1296

It's 5:1 for 1 or less, it should be even (translated to d20 1,2,3,4,5 - critical fail, 20 critical success)...
Not really, make 1 not auto-fail and you're golden.

Anyway, it doesn't really need to be even - it's a choice given to players. If a player wants to, they can use a d20 and have the usual 5% chances. If the same player wants to, they can use 4d6-4 and get a lower chance of rolling low, in exchange for getting a lower chance of rolling high.
Basically, 4d6-4 is what you roll when you want to take 10, but can't.

Kaww
2010-10-09, 11:54 AM
:smallredface: Didn't see that players had a choice... :smallredface:

dobu
2010-10-09, 12:22 PM
Averages.

4d6 = 14 average.
4d6-4 = 10 average.
1d20 = 10.5 average.

0.5 is slight enough.

nope. expected value. not the same as the average.

Eloel
2010-10-09, 12:41 PM
nope. expected value. not the same as the average.

It's the average. 10.5 is not an expected value. It wouldn't be if you had 50% chance at 10 and 50% chance at 11. You have 0% chance of getting 10.5, thus, not expected.

lsfreak
2010-10-09, 12:45 PM
The biggest thing is you really need to think about flat modifiers, because they become incredibly important. A monster that needs a 15 rolled to hit is no longer hit 1:4 but 1:10. The -2 penalty to hit from being sickened can be devastating, while the +2 from flanking can make a near-impossible encounter merely challenging. The DM will need to make sure both the player's abilities and the monster's abilities are in line with each other much more carefully, and possibly nerfing flat +modifiers to d20/3d6 rolls to make sure certain things don't get out of hand. For examples there's a few maneuvers that give +4hit or -4hit, because such a high number is the only way to make it noticeable when you're rolling a d20. With 3d6, flat modifiers are so important that you could probably drop those down to +/-2 and they'd still be really good.

jinete
2010-10-09, 01:21 PM
The biggest thing is you really need to think about flat modifiers, because they become incredibly important. A monster that needs a 15 rolled to hit is no longer hit 1:4 but 1:10. The -2 penalty to hit from being sickened can be devastating, while the +2 from flanking can make a near-impossible encounter merely challenging. The DM will need to make sure both the player's abilities and the monster's abilities are in line with each other much more carefully, and possibly nerfing flat +modifiers to d20/3d6 rolls to make sure certain things don't get out of hand. For examples there's a few maneuvers that give +4hit or -4hit, because such a high number is the only way to make it noticeable when you're rolling a d20. With 3d6, flat modifiers are so important that you could probably drop those down to +/-2 and they'd still be really good.

Huh, it seems there's more game impact with this variant than just more average rolls. I'll be sure to run this by my DM.

Fax Celestis
2010-10-09, 01:32 PM
Averages.

4d6 = 14 average.
4d6-4 = 10 average.
1d20 = 10.5 average.

0.5 is slight enough.

http://img.photobucket.com/albums/v216/FaxCelestis/bellCurve1d20.png
As seen here, d20 has an equal distribution of probability. You are as likely to roll a 1 as you are a 10 as you are a 20. This is a linear/uniform distribution. Any value has ~5% chance to show.

http://img.photobucket.com/albums/v216/FaxCelestis/bellCurve3d6-1.png
3d6 has a bell curve distribution: you are more likely to roll average than you are to roll an outlier. You have ~12% to roll a 10 or 11, and ~1% to roll a 3 or 18. Almost a quarter of the time, you will roll average values. Almost 2/5ths of the time, you will roll between 9 and 12.

http://img.photobucket.com/albums/v216/FaxCelestis/bellCurve4d6-4.png
4d6-4 has a steeper bell curve distribution: you are more likely to roll average than you are to roll an outlier, even compared to 3d6. You have ~11% to roll 9, 10, or 11. Almost a third of the time, you will roll average values. More than half the time (54%), you will roll between 8 and 12.

So no, it is not 'slight'.

mregecko
2010-10-09, 01:39 PM
...numbers...


... I might be in love with you.

:smallredface:

Kaww
2010-10-09, 01:57 PM
@ Fax Celestis

Swordsaged...

Didn't have access to Matlab at the time, wanted to make those right now...

Fax Celestis
2010-10-09, 01:58 PM
Didn't have access to Matlab at the time, wanted to make those right now...

Smallroller (http://www.fnordistan.com/smallroller.html) is smaller and quicker, in my experience.

Kaww
2010-10-09, 02:00 PM
Useful link! Thanks! :smallsmile:

Eloel
2010-10-09, 02:01 PM
So no, it is not 'slight'.
It's a slight change in the numeric average. I have no idea what bell curve and stuff has to do with what the average is.

It might be that English is my 2nd language and we're talking about different things though.
My understanding of average is:
You add all possible values together, divide by the number of values, you find the average.

Fax Celestis
2010-10-09, 02:03 PM
It's a slight change in the numeric average. I have no idea what bell curve and stuff has to do with what the average is.

It might be that English is my 2nd language and we're talking about different things though.
My understanding of average is:
You add all possible values together, divide by the number of values, you find the average.

In a game like D&D, though, outlier numbers matter, not the averages, considering rolling an outlier makes you automatically succeed (20) or fail (1). There is a large, dynamic effect on the game by trading a 5% chance (1d20) for a 1% chance (3d6) or a <1% chance (4d6-4).

Eloel
2010-10-09, 02:07 PM
In a game like D&D, though, outlier numbers matter, not the averages, considering rolling an outlier makes you automatically succeed (20) or fail (1). There is a large, dynamic effect on the game by trading a 5% chance (1d20) for a 1% chance (3d6) or a <1% chance (4d6-4).

I do not deny any of that. Yes, there's a huge difference with outliers. Yes, it tends to give results closer to average more often.

But I haven't ever said it was better/worse/no-effect.

In fact

By calculations, 4d6-4 averages slightly lower, has a chance to roll a 0, but is more consistent around 10. Sounded fair to me.
4d6-4 averages lower. Fact, 10 vs 10.5
Has a chance to roll 0. Fact, even if low chance.
More consistent around 10. Bell curve, as you just said.
Sounds fair. Opinion, can't be proven true or false.

Fax Celestis
2010-10-09, 02:14 PM
I do not deny any of that. Yes, there's a huge difference with outliers. Yes, it tends to give results closer to average more often.

But I haven't ever said it was better/worse/no-effect.

In fact
4d6-4 averages lower. Fact, 10 vs 10.5
Has a chance to roll 0. Fact, even if low chance.
More consistent around 10. Bell curve, as you just said.
Sounds fair. Opinion, can't be proven true or false.

The way you said it initially made it sound like you had no idea about the impact it would have because it seemed (again, from how you said it) that 'if the average is the same, then the distribution must be the same too'.

If you use it, good for you. I'd alter the crit table so that crit threat ranges line up with their expected % rate, instead of basically nerfing them into nonexistence.

Eloel
2010-10-09, 02:18 PM
The way you said it initially made it sound like you had no idea about the impact it would have. If you use it, good for you. I'd alter the crit table so that crit threat ranges line up with their expected % rate, instead of basically nerfing them into nonexistence.

Again, and we'd better get back on 3d6 soon :smallsmile:
It's an option. If anyone wants to take a shot at rolling higher (to crit, to get high on a skill they would fail with an average roll, any reason) (with the drawback of the increased chance to roll lower), they're welcome to roll a d20. Options that are situationally useful but not strictly better are always good, aren't they?

Greenish
2010-10-09, 02:19 PM
If you use it, good for you. I'd alter the crit table so that crit threat ranges line up with their expected % rate, instead of basically nerfing them into nonexistence.He's basically using it as type of "take 10".

Sliver
2010-10-09, 02:33 PM
Again, and we'd better get back on 3d6 soon :smallsmile:
It's an option. If anyone wants to take a shot at rolling higher (to crit, to get high on a skill they would fail with an average roll, any reason) (with the drawback of the increased chance to roll lower), they're welcome to roll a d20. Options that are situationally useful but not strictly better are always good, aren't they?

There is a reason why in the rules you can't always opt to take 10. The option to decide if you want to roll normally or go with 4d6-4, thus increasing your chance of an average roll greatly, undermine this rule. If you go one way or another, this is alright, but letting your players choose reduces risk greatly.

If a player has an average chance of success, he has a good chance of failing on a d20, while not so much on a 4d6-4.

If the player has a high chance of failing, making him require a high roll, on a d20 he has an ok chance (not compared to previous example) while on a 4d6-4 he has almost non.

Letting your players decide every time ensures that most players will opt for the better result, making the risk significantly less.

Kaww
2010-10-09, 02:34 PM
He's basically using it as type of "take 10".

@ FC: Yeah, that's why I made little red faces earlier... Missed it too...

GoodbyeSoberDay
2010-10-09, 03:48 PM
I agree with Sliver. Giving players both options can increase their success rate significantly, especially when it comes to certain would-be-dramatic saves, like facing a Cockatrice.

Oh, I guess this has been explained, but the nasty thing about 3d6 is that high AC opponents are neigh-invulnerable juggernauts to the parts of the party that can't access saves or touch AC. So, yeah, certain encounters will nerf the fighter even more.

A compromise might be to implement 3d6 on skill checks but not attacks or saves.

Dr.Epic
2010-10-09, 04:05 PM
Wait, Crits happen more often. Remember Longswords are max -1, scimitars are -2.
So 18 = Crit with greataxe, 17 with longsword, and 16 with scimitars.

Easier to roll as 16 on a 3d6 than a 18 on a 1d20.

Yeah, that is a good question. And what about improved critical and keen? Not to mention getting a crit with 3d6 (all 6s) is more rare (1 in 216) so shouldn't the crit modifier be higher or something.

Greenish
2010-10-09, 04:08 PM
And what about improved critical and keen? Not to mention getting a crit with 3d6 (all 6s) is more rare (1 in 216) so shouldn't the crit modifier be higher or something.Someone should've thought about that…

Oh, right:
{table]Old Threat Range | New Threat
20 | 16-18
19-20 | 15-18
18-20 | 14-18
17-20 | 14-18
15-20 | 13-18[/table]
http://www.d20srd.org/srd/variant/adventuring/bellCurveRolls.htm

dobu
2010-10-09, 04:23 PM
It's the average. 10.5 is not an expected value. It wouldn't be if you had 50% chance at 10 and 50% chance at 11. You have 0% chance of getting 10.5, thus, not expected.

go look up 'expected value' or 'first moment'.

We're talking about random variables here. ;-) :-)

Eloel
2010-10-09, 04:31 PM
go look up 'expected value' or 'first moment'.

We're talking about random variables here. ;-) :-)

As I said, language barrier, I'm sorry for that.

Knaight
2010-10-09, 05:03 PM
It might be that English is my 2nd language and we're talking about different things though.
My understanding of average is:
You add all possible values together, divide by the number of values, you find the average.
You have to take the frequency of the values into account however. Lets say there are 3 possible values, 1, 2, and 3. The average is not necessarily 2, it depends on how likely each value is. If each of them has a 1/3 chance of happening, then you have 1(1/3)+2(1/3)+3(1/3), which is 2, which is to be expected with 1d3.

Now, lets say you have 1d6, values 3+ count as 3. In that case, you have 1(1/6)+2(1/6)+3(2/3), which has an average of 2.5.

Fax's bell curves on different 1-20ish rolls shows everything necessary. The big changes with less linear distributions are in the value of flat modifiers, as stated prior, by several posters. It should work just fine.