PDA

View Full Version : Integrating Advantage/Disadvantage in a different dice system



Draz74
2018-08-08, 04:20 PM
I'm thinking of picking up my homebrew system again after a couple of dormant years.

To make all the bonuses and penalties in the system matter more, I kind of like the idea of using a bell-curve die roll distribution. The one I'm leaning towards is "roll 3d20, take the middle result." (Anydice demo (https://anydice.com/program/ba6))

But I also mostly like the Advantage/Disadvantage system in 5e, so I'm asking for ideas about how to integrate it with the 3d20-mid system.

If you know beforehand that you're going to have Advantage/Disadvantage, should I just have people roll 2d20 and take the higher or lower as usual? Skip the whole 3d20 business? Or should I have them roll 3d20 take highest/lowest or whatever? I know this is a subjective question, but I'm thinking taking the highest of 3d20 might be too generous, and lowest of 3d20 too harsh.

The tricky part comes when I want to apply Advantage retroactively to a roll, similar to how Lucky works in 5e.
- Should I have them roll 3d20 as normal, then 3d20 again and take the better middle result?
- Should I have them roll 3d20 as normal, then 1d20 again and take that result if it's better than the previous middle result?
- Should I have them roll 3d20 as normal, then 2d20 take highest and take the better result?
- Or should I figure that since applying Advantage retroactively is a generous mechanic anyway, and let them roll 3d20, look at the results, and then decide (at some cost) to apply the high roll instead of the middle roll?
- Or something else?

Again, subjective. But what would you want to play? Keep in mind that I'm expecting this to be played mostly on Roll20, where the original 3d20-mid mechanic can be automated rather than people having to actually roll 3 dice and think about which one to keep.

Mark Hall
2018-08-08, 04:27 PM
With your middle-20 system, I think you'd integrate advantage and disadvantage very smoothly.

Standard? Take the middle die.
Advantage? Take the best die.
Disadvantage? Take the worst die.

Disadvantage gets pretty nasty in this situation, but it's a clean use of the mechanic.

Haldir
2018-08-08, 05:07 PM
I am curious about this question, but from the perspective of smaller dice rolled more often.

Is there any strong feeling about it when you're throwing a bunch of categories of D6's?

Jama7301
2018-08-08, 05:49 PM
I am curious about this question, but from the perspective of smaller dice rolled more often.

Is there any strong feeling about it when you're throwing a bunch of categories of D6's?

If it's a dice pool, you may be better off adding or subtracting dice, or increasing/decreasing success ranges (maybe for d10s on that one).

I think some Powered by the Apocalypse systems go from 2d6 to 3d6, take highest/lowest 2 for Advantage/Disadvantage.

Reversefigure4
2018-08-08, 06:34 PM
With your middle-20 system, I think you'd integrate advantage and disadvantage very smoothly.

Standard? Take the middle die.
Advantage? Take the best die.
Disadvantage? Take the worst die.

Disadvantage gets pretty nasty in this situation, but it's a clean use of the mechanic.

Bingo Bango - precise, clean, and simple. I don't think you'll beat it. The only question is whether you're happy with the relatively larger skew in probabilities (vs 5e's 2d20 dis/advantage mechanic). Advantage is much better with 3d20, and disadvantage is much worse.

Personally, I'd steer away from it, since it moves from "bonuses and penalties matter more" to "getting advantage or avoiding disadvantage are critical", and the dice roll matters a lot more (defeating the point of 3d20 in the first place).

Knaight
2018-08-08, 07:45 PM
I'd add a fourth die instead of going to two, then use the second best or second worst. This bumps the mean by 2 either way, but makes the most likely results 7 and 14, with high probability results clustered around them.

As for the bunch of d6 mentioned later in this thread, that sounds like a roll and keep system which uses unusual d6 dice (Genesys, maybe?), or just multiple color coded d6 at once for different things (e.g. Mythender, but I don't have a specific guess here). I'd look into how roll and keep systems operate to figure out how this one should operate.

Draz74
2018-08-09, 12:03 AM
I'd add a fourth die instead of going to two, then use the second best or second worst. This bumps the mean by 2 either way, but makes the most likely results 7 and 14, with high probability results clustered around them.

That's a very interesting possibility that I hadn't considered.

Draz74
2018-08-09, 12:35 AM
I'm not sure I like it, the more I think about it. For one thing, taking the 2nd-highest die of 4 means it's very likely to be better than average, yes, but unlikely to be spectacularly good either. Normal Advantage has the exciting relative probability of getting a 20!

kieza
2018-08-09, 01:03 AM
The problem with 3d20, keep middle, is that it doesn't actually produce a bell curve. (The distribution has the "hump" in the middle, but doesn't have the tapered ends.) I'm away from my stats references at the moment, but I think the easiest way to get an actual bell curve is to keep at least 2 dice.

I'm working on a system that does this by keeping 2d10.

A normal roll is 2d10 + bonus, which also isn't quite a bell curve, but becomes one when modified by an advantage/disadvantage mechanism:
With "favor," you roll one extra die and drop the lowest.
With "disfavor," you roll one extra die and drop the highest.
If you have both favor and disfavor, they stack: roll 4d10, drop both the highest and lowest. This results in a probability distribution which de-emphasizes extreme values, but has the same range and mean as 2d10.

A roll with either favor or disfavor actually has the classic bell-shaped probability distribution, although it may be skewed to one side.

The point of doing this is to exploit bell curves so that small (+1) bonuses are important when you have a ~50% chance of success, but they have diminishing returns:
When you hit on an average roll of 11, getting an extra +1 bonus moves you from 55% to 64% chance of success. (The bonus is useful 9% of the time.)
When you hit on a roll of 6, getting an extra +1 bonus moves you from 90% to 94%. (Useful 4% of the time.)

Favor amplifies this:
If you hit on 11 and have favor, then the +1 bonus moves you from 78.5% to 84.6%. (Useful 6.1% of the time.)
If you hit on 6 and have favor, then it moves you from 97.7% to 98.9%. (Useful 1.2% of the time.)

Part of the design philosophy in this system is that numbers shouldn't get much larger than they are at level 1. No escalating attack bonuses, hit points, etc. that make mid-level characters into superheroes, although you can raise certain values slightly if you choose to do so. The alternative to raising those values is getting to do cool stuff, though, like apply additional elemental effects to magic, or specialize in a fighting style that grants benefits beyond the strictly numerical. (I think this is also better than how DnD bounded accuracy works: higher level creatures are not significantly harder to hit, and do not have significantly more hit points, but they get to do more stuff per turn: more passive effects, more interactions between abilities, etc.)

Another part of the design philosophy is cutting down on situational bonuses that are "handed out," like a spell that gives a +1 bonus to a specific type of roll. Those are hard to keep track of when you're getting them from another player, and it leads to a lot of slow-downs as people try to make sure that they've counted all their bonuses. So, any bonuses should be one of a handful of categories:
--Always on, and affects everyone in the group so that people get used to them and remember to count the bonus.
--Big and memorable, so that it sticks in the memory when given.
--Only affects the person granting the bonus, who is presumably going to remember about it because they chose it for their character.

Any other beneficial effect should be handled with favor. With some thought given to tactics, you can get favor often enough that attack bonuses (for example) rapidly encounter diminishing returns.

Knaight
2018-08-09, 01:29 AM
I'm not sure I like it, the more I think about it. For one thing, taking the 2nd-highest die of 4 means it's very likely to be better than average, yes, but unlikely to be spectacularly good either. Normal Advantage has the exciting relative probability of getting a 20!
I figured that was intentionally being avoided - hence the 3d20 to start with. That said, the odds of a 20 do still increase with 4d20, albeit from 0.72% to 1.40%. Obviously seeing how it works in play is most important, but looking at the curves in Anydice it is really pretty. Plus using something like 18+ where you'd use 20 normally (if you want to keep the crit analog) produces a much larger bump, from 6.07% to 10.95%.

My main concern is the time added to find the second highest die. 3d20 find middle is already not that fast, and going from three to four dice at the table can cause a surprising amount of slow down. I don't know how rolling intensive your system is, but if you want something like routine iterative attacks it could get really gross really quickly.


The problem with 3d20, keep middle, is that it doesn't actually produce a bell curve. (The distribution has the "hump" in the middle, but doesn't have the tapered ends.) I'm away from my stats references at the moment, but I think the easiest way to get an actual bell curve is to keep at least 2 dice.

I'm not sure the bell curve is really necessary though - a clean arc seems to work just fine for the purposes here, as would the classic two die chevron or plateau chevron (though if you do drop dice you start getting that bell curve, middle 2 of 4d20 gets a very classic bell curve shape). It's the straight line probability distribution which is being avoided.

Draz74
2018-08-09, 07:51 AM
The problem with 3d20, keep middle, is that it doesn't actually produce a bell curve. (The distribution has the "hump" in the middle, but doesn't have the tapered ends.) I'm away from my stats references at the moment, but I think the easiest way to get an actual bell curve is to keep at least 2 dice.
I think you need at least 3 dice to get an actual bell-shape. That said, I don't think I care about having the tapered ends, like Knaight said. I think I'm fine with the distribution shape ("chevron"???) of 3d20-mid.


I'm working on a system that does this by keeping 2d10.

A normal roll is 2d10 + bonus, which also isn't quite a bell curve, but becomes one when modified by an advantage/disadvantage mechanism:
With "favor," you roll one extra die and drop the lowest.
With "disfavor," you roll one extra die and drop the highest.
If you have both favor and disfavor, they stack: roll 4d10, drop both the highest and lowest. This results in a probability distribution which de-emphasizes extreme values, but has the same range and mean as 2d10.
That last bit (favor AND disfavor) is interesting, although I actually think I like the flavor of rolls becoming MORE swingy, not less swingy (more reliable), when modified by both Advantage and Disadvantage. In my system, if you have both, I might have that reduce to just a straight d20 roll. It's the opposite of what you're doing, but thanks for the idea anyway!


A roll with either favor or disfavor actually has the classic bell-shaped probability distribution, although it may be skewed to one side.
Good to know.


Part of the design philosophy in this system is that numbers shouldn't get much larger than they are at level 1. No escalating attack bonuses, hit points, etc. that make mid-level characters into superheroes, although you can raise certain values slightly if you choose to do so. The alternative to raising those values is getting to do cool stuff, though, like apply additional elemental effects to magic, or specialize in a fighting style that grants benefits beyond the strictly numerical. (I think this is also better than how DnD bounded accuracy works: higher level creatures are not significantly harder to hit, and do not have significantly more hit points, but they get to do more stuff per turn: more passive effects, more interactions between abilities, etc.)
Yep, that's the goal for me as well. Well, leveling up does get you SOME increase in bonuses over a Level 1 character in my system, but nothing like in 3.5e. (For example, a high-level character can pretty easily get a +10 bonus to a Skill they specialize in, compared to a +3 for a Level 1 character.)


Another part of the design philosophy is cutting down on situational bonuses that are "handed out," like a spell that gives a +1 bonus to a specific type of roll. Those are hard to keep track of when you're getting them from another player, and it leads to a lot of slow-downs as people try to make sure that they've counted all their bonuses. So, any bonuses should be one of a handful of categories:
--Always on, and affects everyone in the group so that people get used to them and remember to count the bonus.
--Big and memorable, so that it sticks in the memory when given.
--Only affects the person granting the bonus, who is presumably going to remember about it because they chose it for their character.
Absolutely, a big part of the goal here is to cut down on temporary/situational bonuses -- "handed out," as you say. In fact, I don't even make exceptions for "big and memorable" or "only affects the person using the ability" effects.

I do make exceptions for 3 cases that are built into the system, not dependent on any spell or other special ability, which should be common enough that players get used to having to apply them:
A -2 penalty to all saves when a character is Wounded.
A +2 bonus to attack Accuracy when a character has "combat advantage" (most commonly from Flanking, but also if the target is Dazed or something).
A scaling -2/-5/-10/-20 penalty to attack Accuracy when attacking from ranges longer than the ranged weapon was intended for. This one gets a little messy, but my simulationist side hasn't found another solution that I like yet.
Other temporary/situational bonuses can be represented by other die mechanics. Advantage/Disadvantage, or "Coasting" (getting to assume a minimum die result, like Reliable Talent in 5e), or replacing one statistic on your character sheet with another statistic from your character sheet.


I figured that was intentionally being avoided - hence the 3d20 to start with. That said, the odds of a 20 do still increase with 4d20, albeit from 0.72% to 1.40%. Obviously seeing how it works in play is most important, but looking at the curves in Anydice it is really pretty. Plus using something like 18+ where you'd use 20 normally (if you want to keep the crit analog) produces a much larger bump, from 6.07% to 10.95%.
Hmmm. Could you please link the Anydice distribution? (I'm not actually much good at using the site.) Now that I think about it, I'm not sure my system really cares much about nat 20's anyway ... so I shouldn't have been worrying about that.


My main concern is the time added to find the second highest die. 3d20 find middle is already not that fast, and going from three to four dice at the table can cause a surprising amount of slow down. I don't know how rolling intensive your system is, but if you want something like routine iterative attacks it could get really gross really quickly.

No my system definitely does NOT have routine iterative attacks, but still definitely a valid concern. Particularly since I'm pretty sure there's no way to automate it in Roll20. (Roll20 may be able to do 4d20-take-second-best, but not if you roll the first 3d20 and the last d20 separately.)

Like I said in the OP, I may just keep standard Advantage/Disadvantage (2 dice) distributions when a "reroll" is warranted ahead of time. When a player has already rolled 3d20 and wants to apply a "reroll" retroactively, though, is when I may use your 4d20-take-second-best idea. And maybe special abilities that do that will be rare enough that the slowdown will be negligible.

JBPuffin
2018-08-09, 08:45 AM
Drop to 2d20 with Adv/Disad, keep as appropriate. Drop to 1d20 when you have both. Maybe also saying something like “you can’t roll less than 7 with Adv or more than 14 with Disad”?

Knaight
2018-08-09, 08:51 AM
I think you need at least 3 dice to get an actual bell-shape. That said, I don't think I care about having the tapered ends, like Knaight said. I think I'm fine with the distribution shape ("chevron"???) of 3d20-mid.
That shape was the arc. Something like 2d10 would be a chevron. 1d8+1d12 would be a chevron plateau. You also need to roll at least 3 dice, middle 2 of four absolutely produces a bell curve. I'll throw that in with the Anydice at the end of the curve.


Hmmm. Could you please link the Anydice distribution? (I'm not actually much good at using the site.) Now that I think about it, I'm not sure my system really cares much about nat 20's anyway ... so I shouldn't have been worrying about that.

No my system definitely does NOT have routine iterative attacks, but still definitely a valid concern. Particularly since I'm pretty sure there's no way to automate it in Roll20. (Roll20 may be able to do 4d20-take-second-best, but not if you roll the first 3d20 and the last d20 separately.)

Like I said in the OP, I may just keep standard Advantage/Disadvantage (2 dice) distributions when a "reroll" is warranted ahead of time. When a player has already rolled 3d20 and wants to apply a "reroll" retroactively, though, is when I may use your 4d20-take-second-best idea. And maybe special abilities that do that will be rare enough that the slowdown will be negligible.
Here you go.


output 2@4d20
output 2@3d20
output 3@4d20
output [middle 2 of 4d10]


Those are four different commands, the last of which is just to demonstrate the 2 dice out of 4 curve.

Draz74
2018-08-09, 09:35 AM
Thanks for the instructions.

Anonymouswizard
2018-08-09, 10:23 AM
Yeah, the main problem is the specific die rolling system. In almost anything other than a dice pool you can add in advantage quite easily (add additional die, ignore best/worst), but 3d20 take the middle is making it much more difficult to integrate smoothly.

The other one that you might have trouble with is d% systems. Although in cases of massive advantage/disadvantage I'd force the player to use the highest/lowest result instead of designating dice as tens and digits (which in 'roll high but not over' systems has the fun consequence of making it so they'll tend to give a relatively minor success or extreme failure).

Haldir
2018-08-09, 04:33 PM
Yeah, the main problem is the specific die rolling system. In almost anything other than a dice pool you can add in advantage quite easily (add additional die, ignore best/worst), but 3d20 take the middle is making it much more difficult to integrate smoothly.

The other one that you might have trouble with is d% systems. Although in cases of massive advantage/disadvantage I'd force the player to use the highest/lowest result instead of designating dice as tens and digits (which in 'roll high but not over' systems has the fun consequence of making it so they'll tend to give a relatively minor success or extreme failure).

So, if I am working with a game that wants to generate a D6 roll => a listed number, with three dice. Each dice must turn up that number or greater...

IF D=4
AND D1 => 4 \\\ D2 => 4 \\ D3 => 4
THEN
Attack proceeds.

How do you guys feel about limiting or granting advantage based on another mechanic or GM judgement?

Mark Hall
2018-08-10, 10:14 AM
Yeah, the main problem is the specific die rolling system. In almost anything other than a dice pool you can add in advantage quite easily (add additional die, ignore best/worst), but 3d20 take the middle is making it much more difficult to integrate smoothly.

The other one that you might have trouble with is d% systems. Although in cases of massive advantage/disadvantage I'd force the player to use the highest/lowest result instead of designating dice as tens and digits (which in 'roll high but not over' systems has the fun consequence of making it so they'll tend to give a relatively minor success or extreme failure).

I think advantage giving you the best d20 and disadvantage giving you the worst is a pretty smooth implementation... the math isn't the same, but it is a very simple and evocative use of the idea.

For % systems? Standard is designate a die. Advantage is "Best interpretation". Disadvantage is "worst interpretation". Again, the math isn't the same, but its a smooth integration.

Max_Killjoy
2018-08-10, 11:41 AM
For a system using 3d6, you could simply add a die either way, keep the worst 3 with disadvantage, keep the best 3 with advantage.

PinkSpray
2018-08-10, 08:22 PM
You're talking math here.

I can tell you from running 5e for groups of vets & beginners: Advantage/Disadvantage works magic :smallbiggrin:

Players react to taking the higher roll very well - especially if they rolled low initially. It lights the whole table up.

The whole concept was drawn from "at the table excitement" push that's part of gaming logic. I gauged the players
reaction very similar to rolling a nat 20. It energizes the whole group, and even when the high roll failed, there was
a strong air of anticipation at key moments that mattered. From a gaming perspective.

The first system I considered adding it to was GURPS, since it can be heavy with situational modifiers. Should
work but I've yet to playtest it. "Roll 3d6 twice" is definitely easier. I'll see. Hell, trying it with Pathfinder would
be interesting.

It's real gaming innovation :smallwink:

Knaight
2018-08-11, 01:02 AM
It's real gaming innovation :smallwink:

Roll and Keep was a pretty impressive bit of design back when it was developed in 1995 for the first edition of L5R (though there might be a predecessor I don't know about). I'd call it innovative.