METTLE Hybrid Dice Pool

There is a recent trend towards RPG designers leaving design notes in their actual game books. I have mostly avoided them in my work. If the reader cares they will find out soon enough through play, and if they do not it was probably unimportant. Perhaps the readers will make a video or write a blog post about the "unspoken design intent" or some such - why spoil their fun by explaining it before they do?

     In METTLE Core I had one lapse. Gaze upon this one shameless morsel of designer narcissism, from the Engine chapter:

Note: the average score of a single die is 1, making it easy to gauge chances of success. This also means you can directly contest single attributes, using one as the active check and the other as the passive difficulty.

     I like to think this was restrained, driven by the obvious necessity of explaining why I just didn't use a "normal" dice pool everyone is already familiar with. In this largely unread blog well outside of my similarly unread game books, I am less constrained by good taste and dignity.

     So let's dive deep into why I used a strange hybrid dice pool for METTLE!

Direct Contests

The biggest reason is that it allows you to directly Check a single Attribute against another, rolling one Attribute and using the other as the Difficulty. In METTLE they are equivalent! This may seem abstract but it really makes the design space much more flexible.

     The usual solution in other dice pool systems is opposed Checks or "Contests". Here both sides roll their die pool. This is also direct, but eats up time and patience for both the player and game master - especially if there are a lot of enemies or actions.

     Another typical solution is to set a defender's stat as the difficulty of a check, but with standard one die equals one success pools, the difficulty is too high! Most dice pool outcomes average out at about half of the pool or less. Other games that do this get around this with an indirect kludge: halve the defending pool or add an Attribute or Skill to the attacking pool. This pumps the number of dice up pretty high and the effect of the doubled or halved pool will scale oddly as numbers increase.

Margins/Edge

Most dice pool systems count the number of successes over the difficulty as a margin-of-success. In METTLE there is no subtraction, you just count your Edge (4-6 faces). Edge dice would normally just be failures (or worse, "scoundrels") in other dice pool systems, but here every die that falls is important. You want to roll 1-3s for a good Score, but you also want 4-6s for Edge!

Probabilities

The central conceit here is that the average score on any die is 1, ranging from 0 to 3. You can check this yourself: adding the Scoring faces and dividing by the total faces gives us:

 (1+2+3) / 6 = 1

     It's also a pretty minor feat in Anydice to check this: 

Mean/average is always the same as the number of dice, deviation increases reliably with dice, and the maximum is always 3x the pool. Really a lot of desirable behavior.

Origins

This traces back pretty far to a post on RPG.net back when you were allowed to argue about game mechanics on the game design section. Believe it or not, what is now a silent desert was once a thriving little ecosystem of horrible nerds.

https://forum.rpg.net/index.php?threads/d6-system-but-reading-the-dice-differently.782505/post-20113769

They did bring up a fair criticism that the modal value of the dice was a little odd at lower dice pools (<5), but this is also a reason I avoid using die penalties and raise Difficulty instead.

Conclusion

That's all for now. I'm probably not posting this widely because it is very niche as far as interest.

Pipe up if you have any comments. You know what to do.

 

  

D&D B-Series Pregen Ability Scores

I have been ruminating about old school D&D Ability Scores, and whether they meant all that much. So I decided to run the numbers. I entered 146 pregenerated character stat lines found in the B series of adventure modules into a spreadsheet and imported them into R Studio. Pooped out some descriptives and ridgeline plots, and here is the result. It's more of a descriptive analysis than an inferential one - I do plenty of the latter in my day job.

 

Keep in mind that I can only look at what the designers chose for their Basic modules. I was active in the 80s and the actual state of play was all over the place. Overpowered characters with ludicrously high scores were ever-present. I propose that the more restrained arrays in the modules are reflective of designer intent, whatever you believe that is worth.

What I wanted to find out:

• How did the carefully pregenerated module characters stack up to 3d6 down the line?
• What would an average representative of a Basic D&D Class look like score-wise?
  

Overall Ability Scores by Class

Ideally, Character creation in Basic D&D involves rolling up a set of six Ability scores from 3-18 each, then seeing what sort of Class they qualify for. A Magic-user needs an INT higher than 9, etc. This first plot is the average Ability score for a given class, lumped together. This gives a broad overview of what the actual average score is and which classes tend to get assigned better rolls. At this point there were no Ability score bonuses for Races, so everything is conveniently 3-18.

 

To show this I use ridgeline plots, a kind of smoothed aesthetically-pleasing histogram. I have added a dotted vertical line in the middle of each to show the mean (10.5) for a 3d6 roll. The color also shifts with the score - lower scores are bluish, higher scores reddish.

Well, would you look at that - an awfully suspicious trend towards better-than-average! Is this a selection bias to ensure that players get decent pregens, or a result of Moldvay's option to reroll "hopeless" characters? 

 

The world may never know. 

 

The average module pregen actually sits at 11.4, or roughly 11-12. This is a slight boost over the 10-11 expected with purist 3d6 rolls. Sorry, but I can't be bothered to check for statistical significance, even with R open as I am writing. Maybe later.

Other take-aways:

• Clerics and Thieves seem to get assigned the worst overall rolls.
• Elves and Dwarves seem to get assigned the best overall rolls. 
  
• Magic-users and Dwarves seem to occupy more of the very high results, shown by the taller and longer red tails.
• Halflings have a weird small bump in the low/blue end, perhaps a population of crappy Halflings that throw their otherwise good curve off.
• Fighters are very consistent, almost a normal curve for their rolls.
  

Classes Breakdown

What follows is a little different: the average score for each individual ability, separated by Class. This shows what specific ability scores are favored or disfavored when considering a Class. Sometimes, the histogram splits into two peaks, and it is debatable what that means.

 

To preface some things I will mention later, Basic D&D also had a thing called a Prime Requisite, which is an Ability Score or two that the designers deemed important for that Class. If they were good (13+) you would get a bonus to earned XP. If they were bad, you could get a hefty penalty to XP. I believe these explain most of the patterns you see below.

 

(Images by Terry Dykstra, from the Rules Cyclopedia)

Middling folks relying on their connection to a (usually) unspecified deity! Rolled arrays with a high WIS and little else going for them seem to become Clerics. WIS is entirely for the Prime Requisite XP bonus I think. It only enhances magic saves, which everyone else needs just as much. Clerics tend toward lower INT, probably because religion is silly. To be fair though, the power and versatility of their Class more than makes up for the quality of its members.

The module authors are really giving some good scores to Dwarves. STR and CON are reliably huge. INT, CHA, and DEX tend to suffer, but the two peaks suggest two different camps on whether their WIS and CHA should be high or low. The stereotypes weren't quite set in stone then perhaps.

 

As a special note, see the low blue tail on CON that shouldn't be there. This is partly the ridgeline smoothing extending the curve, but also some mistaken roll assignments. Dwarves have a minimum CON of 9, but tell that to the module authors. This happens elsewhere too (see Halflings).

The best scores reliably go to Jacked-Genius Elves. If you get an array that could make one, I guess you just go for for it. The peaks might show a little division on whether their CON and WIS scores should be low or high.

Halflings were oddly beefy for their size, which no one seemed to think twice about. Despite basically being Fighters, they had minimums and prime requisites that meant the best scores would often go to them. This was probably intended to keep them rare, like elves, but also just made most adventuring exemplars of their race pretty outstanding.

As Ash Williams might say: "Hit first, Think never." Good physical scores and low mental ones. There is a broader range of CHA - a high one makes sense if you want them to be a good leader at domain level, a low one if they are just a rough-neckin' thug.

Scraggly nerds, but they are investments for later godliness, so you want to give them a solid set of scores. A high INT and low STR is obvious, everything else is a nudge above average. Note that like the Cleric, their prime requisite of INT is only good for the XP Bonus.

Light-fingered and charming, with little else going for them. Most of the time they will be relying on their special abilities rather than their Scores anyway - and even their godly DEX has no effect on these in Basic.

Summary of Breakdowns

So if you just "need a guy" real quick, you won't be too bad off using one of these arrays:

 

A Word on Prime Requisites

The idea of high Ability Scores primarily just giving an XP bonus to certain Classes may seem strange to a modern player. This is often its largest benefit in B/X. For example, the benefit of having a higher INT is mostly that your Magic-user rises through their levels a bit more quickly. It has no direct effect on their number of spells or spells known. It only has an impact on their magic in a roundabout way, unlike later editions.

A Word on Minimums

Most Classes also had minimum scores, but those probably had less influence. There are even errors there, for example some Halfling pregens had CON scores below 9! I have left similar genuine mistakes in the database uncorrected, only correcting obvious typos like a Fighter with a CON of 1. In that case I changed it to an 11 - "1" is a common typo for a repeating number like 11 and close to the average anyway.

Conclusion

I don't think anyone came away from this with any surprises. Pregenerated characters had slightly higher Ability Scores than 3d6 down-the-line would imply, Ability Scores tended to reflect the Prime Requisite(s) of the Class, Clerics blow, and Elves rule. I had fun doing it though and hope you got something out of it too, even if it was just confirming your suspicions.

Comment below, especially if there is something else you want to look at.