Designing a d20 RNG

[DenizenKane]

Been messing around with converting outputs to wounds, and I've got this Anydice program (it gives me an error when exporting the link).

It lets you experiment easily with the values, Z is the range of each wound increment, Y is the starting mod. I think it's best to have no base DC for soak, that way if you're even with someone you'll do atleast 1 wound if you land a hit. The smaller you make Z, the more swingy things get. It seems ~8 is about ideal, because it's ~2 standard deviations and lets you slide smoothly.

Z:8
Y:0
X:(2d10+Y)
output [count {-100..0} in X] named "No Wounds"
output [count {1..Z} in X] named "1 Wounds"
output [count {Z+1..(2*Z)} in X] named "3 Wounds"
output [count {(2*Z)+1..(3*Z)} in X] named "6 Wounds"
output [count {(3*Z)+1..(4*Z)} in X] named "10 Wounds"
output [count {(4*Z)+1..(5*Z)} in X] named "10+ Wounds"

[DrPraetor]

You're going to write an entire ruleset, presumably, and you want the staging between wound levels (first edition Shadowrun rules terms: I choose you!) to be the same as the staging between degrees of success.

That is, if a result of Y is a minor wound and a result of Y+Z+1 is a moderate wound, then:
on some non-damage check, if a result of Y is a success than a result of Y+Z+1 should be one degree of critical success.

Looking at your anydice script result, you have, for "basic" attack where Damage = Soak and thus Y=0:
28% of 1 box
62% of 3 boxes
10% of 6 boxes

Notably, you need Y=5 to have a 1% of taking your opponent down in a hit (a natural 20) and Y=14 to have a 52% of taking your opponent down in a single hit. So in order to take Orcs out in a single hit reliably you need to be +14? That's neither good nor bad, just be aware that when you are "less swingy" (as you now are, and as I vaguely suggested you should be, so I can hardly criticize) is that you need to push yourself off the RNG in order to take out enemies in a single hit, which idiomatically you will want people to be able to do.

[DenizenKane]

Code: Select all

Z:5
Y:0
V:3
X:((2d10-V)+Y)
output [count {-100..0} in X] named "No Wounds"
output [count {1..Z} in X] named "1 Wound"
output [count {Z+1..(2*Z)} in X] named "3 Wounds"
output [count {(2*Z)+1..(3*Z)} in X] named "6 Wounds"
output [count {(3*Z)+1..(4*Z)} in X] named "10 Wounds"
output [count {(4*Z)+1..(5*Z)} in X] named "10+ Wounds"

I'm liking how this looks. If you go the "Transposed" tab it's easier to read the chances.

EDIT: Link worked: https://anydice.com/program/1156d

[DenizenKane]

https://anydice.com/program/1162d

3d6 gives a better curve, and this program shows a range of modifiers at once.

[deaddmwalking]

Clearly having more dice reduces extreme events and brings things further into the middle. Rolling a 3 (or 18) on 3d6 has a 1 in 216 chance. You do have to ask why you want that, and not, say, 5 dice. With enough dice you'll always get an average result. Too far in that direction and there's no point in rolling.

If you're rolling dice you want variance. Choosing your dice involves deciding how much is appropriate. Every die you add decreases variance to some degree. Of course, it also makes the addition more complicated. Adding 3 numbers is more than adding 2 numbers. The degree of complexity is rather small because the individual numbers are single digit numbers, but it is still more. If you opt for 3d6 instead of 2d10 or instead of 1d20, you have to explain what 'extreme events' justify a less than 1/2 a percent chance compared to a 1% or a 5%.

For myself, I want a slight clustering toward average, but I want extreme results to happen generally frequently. I like having a character 10x more likely to roll their 'average' than to have their best (or worst) day. It isn't wrong to make average 20x or 200x more likely, but you have to remember that 1) the numbers are in service of the game and 2) real people are calculating these results at the table.

[MGuy]

So I'm guessing this thread is no longer about that d20.

[Iduno]

So I'm guessing this thread is no longer about that d20.

Sounds more like replacements for d20 that have a similar feel but better math?

[MGuy]

So I'm guessing this thread is no longer about that d20.

Sounds more like replacements for d20 that have a similar feel but better math?

Wut? 3D6 doesn't have a similar feel to 1d20 either literally or in how it pans out in play. It's a differentl RNG that producers different (not inherently better) numbers. I'm not questioning the desire to experiment with other dice but the thread seems to have drifted significantly away from designing around the d20.

[DenizenKane]

Deaddm: Yeah, I think you have a good point. 2d10 does seem to be the sweet spot when it comes to curved RNGs. 3d6 gives a more clustered curve, but it's harder to calculate and it may be too deterministic.

And, yeah this thread derailed a bit from the title, but things evolve. I don't think a flat RNG works for the game I'm trying to create, because I want characters to be able to consistently succeed at tasks while still having reasons to roll dice.

Anyway, as for the game premise:

It's "DnD" with reworked mechanics. There is heavy focus on combat, but ideally there will be working "minigames" for many parts of the game including social encounters, exploration, economy, dominions, dungeon crawling, and chases.

The game will include mechanics to help facilitate classic RPG adventure stories, possibly following the "Hero's Journey" story structure. Characters will be swept in a "save the world" plot where they have to fight off the big bad evil and the rogues gallery of villains helping him carry out his/her plans for world domination. PC deaths should be rare, but possible.

The main setting will be a deconstruction of kitchen sink high fantasy with a fresh coat of paint. Monsters will be both a combat encounter, and have defined flavor, ecosystems, and environments they live in and create. Care will be taken to make the world "come alive."

Alignment is on a ROYGBP color wheel with your choice giving you small bonuses on things related to your alignment.

Encounters should be expected to be around 30-40 minutes, and last about 4 rounds. Combat encounters will usually include 4+ creatures with different roles to keep things interesting. Mechanics should be streamlined with minimal dice rolling as to keep the game moving.

Potentially the main RNG is going to be 2d10, and HP is fixed to allow things to scale.

The game and it's classes will be split into 4 tiers, Basic classes will run from 1-5, advanced from 6-10, prestige from 11-15, and epic from 16-20. Multiclassing will be accomplished by special multiclass versions of classes that can voltron together.

Classes will each have unique resource management mechanics aimed at making the game more fun to play, but not too fiddly as to be burdensome or too time consuming at the table.

Basic characters will focus on small scale and "gritty" encounters, advanced classes have heavy doses of awesome and magic and have more control over the world around them. Prestige level characters are kings thats control territory and have lots of room for horizontal advancement. Epic level characters are gods that rule of very large domains, potentially whole planes.

[deaddmwalking]

The game will include mechanics to help facilitate classic RPG adventure stories, possibly following the "Hero's Journey" story structure. Characters will be swept in a "save the world" plot where they have to fight off the big bad evil and the rogues gallery of villains helping him carry out his/her plans for world domination. PC deaths should be rare, but possible.

Nothing about your resolution mechanic is going to facilitate The Hero's Journey.

Here's the short version pulled from a chart online:
1) Call to Adventure
2) Meeting a Mentor
3) Crossing Threshhold
4) Trials & Failures
5) Growth, New Skills
6) Death & Rebirth
7) Revelation
8) Finally Changes
9) Atonement
10) Gets Gift
11) Return Changed

For an adventure, you can certainly use a structure like this. Heck, for a series of adventures, you could do this. Luke in A New Hope and Luke in Empire Strikes Back basically follow the same pattern. Or at least, there are epicycles in the journey that repeat.

If you're creating a system, I don't think you want to lock yourself into just one narrative structure. Certainly, if you want people to play it again and again you probably want to avoid having 'save the world' plots over and over again.

Even around combat it sounds like you're too fixated. Having encounters be possible as a solo, pair, ~4, and ~12 would be good. A fight against a wave of mooks won't be memorable if that's the only fight that's possible, and there isn't much difference between fighting one ogre and one minotaur. But changing up the number and composition of enemies will make the fights feel much more different than changing the roles.

You're approaching this project with a lot of unchallenged assumptions. You know you want to do things differently, but it appears you don't understand why you want to do them differently. When you say 'basic tier' and 'advanced tier', that should mean something to you, and you should be able to explain it.

Why does Basic need to be 5 levels? Why not just one level? Is it because of hit point scaling? Is it because of ability acquisition? What should an advanced character be able to do that makes them different than a basic character? Is it just a +1 to Attack/Defense?

[erik]

I recommend getting rid of level numbers entirely. Then people don’t feel compelled to start at 1 which is a shitty place to start since by necessity the lowest run on the ladder should be garbage to simulate those who cannot adventure. DnD got past this some with fractional CR but that’s sloppy.

Also if not using numbers then you aren’t pigeon-holed into a pattern and you can make your tiers as long as needed and no longer.

[Hicks]

It depends on how you handle levels. You can just flat out *make* the character generation start at level 3, and say NPCs have one or two levels. Or level whatever.