Tuesday, March 8, 2022

Skill Roll Math for Old School D&D

TLDR: skill checks should be made only when absolutely necessary and should probably never utilize the roll-under-ability score method. Various alternatives are analyzed.


The following largely pertains to iterations of D&D from 1974 to 1981. That is, in the era when the game did not have a strong concept of skill rolls outside of predefined elements within a class or ability. My baseline for the research below is Old School Essentials, which should be aligned with 1981 Moldvay/Cook Basic and Expert D&D. The purpose here is show how various methods of rolling skill checks stack up against prescribed skill checks by class.

Skill Check Methods Presented in OSE

In particular, I want to start with the fact that OSE offers a number of various ways to do skill checks. In classes, there are both x-in-6 and percentile chances expressed. (I consider a Save, which is d20 based, as something other than a skill check.) On page 216 of the OSE Classic tome, Gavin suggests that when a situation comes up that is not already covered by the rules (e.g. skill rolls as defined by class) the GM can "require the player to make an ability check (see p104) or a saving throw (see p105) to determine what happens. Other times, the referee may judge the likelihood of the action succeeding (e.g. expressed as a percent- age or X-in-6 chance), tell the player the chances, and let them decide whether to take the risk or not." For the record, Gavin describes an ability check as rolling d20 ≤ the relevant ability stat, sometimes with a modifier of up to +/- 4 applied by the GM. In the OSE Advanced Player's Tome, the ability check is the only method given (p. 204) for this purpose, but the Referee's Tome echoes the above paragraph, providing multiple methods.

Skill Check House Rules Survey

On my podcast, Plundergrounds, I recently broached the topic of skill checks and had quite a few suggestions and/or "this is how I handle it" call-ins. Among the suggested methods (aside from those defined by OSE above), were:

  • Roll d100 ≤ ability
  • Roll 3d6 (or even 4d6 or 5d6) ≤ ability
  • Roll d6 literally modified by relevant ability ≥ 6
  • Roll a pool of d6s equal to modifier of relevant ability ≥ 6 (if 0 or less roll 2d6 and take the worst)
Several individuals said they just set x-in-6 chances. (Spoiler alert - that's a pretty good way of doing it.)

Baseline Skills

Surveying level 1 characters in OSE Classic, the following skill areas are defined. Where a chance reads 33.3%, it has been converted from a "2-in-6" designation in the text.
  • Open Doors 17-83% (x-in-6 based on STR)
  • Detect Construction (Dwarf) 33.3% 
  • Detect Room Traps (Dwarf) 33.3% 
  • Listen at Door (Dwarf, Elf, Halfling) 33.3% 
  • Detect Secret Door (Elf) 33.3% 
  • Hide in undergrowth (Halfling) 90% 
  • Hide in dungeon shadows (Halfling) 33.3% 
  • Climb Sheer Surfaces (L1 Thief) 87.0% 
  • Find/Remove Treasure Trap (L1 Thief) 10.0% 
  • Hear Noise (L1 Thief) 33.3% 
  • Hide in Shadows (L1 Thief) 10.0% 
  • Move Silently (L1 Thief) 20.0% 
  • Open Locks (L1 Thief) 15.0% 
  • Pick Pockets (L1 Thief) 20.0%
If you subtract the more specialized Thief skills (in red) from this line-up and acknowledge that the STR-based Open Doors is going to be 33.3% (2-in-6) for the majority of characters (characters with an average ability score), you see the trend that a special class skill usually is set to that baseline of 2-in-6. 

This is an important baseline, because on-the-fly skill checks called for by the GM probably shouldn't supersede these class-based abilities.

Method Math

The following graphs and methods are compared to see which ones are in line with the spirit of the baseline skills of OSE. 

Dataset 1: showing d20 and 100 rolls vs. ability score target numbers

The Ability Check (row 3)

OSE recommended d20 ability checks are a bad idea! In one sentence: an ability check against an ability score of 7 (low by any one's standards) is superior to a baseline 2-in-6 chance (35% vs. 33.3% respectively). 

Let that sink in a minute. A character with an ability score of 7, well below average, will be more likely to succeed at a skill check than if that same character had a prescribed 2-in-6 class ability printed on their character sheet. Think of how this applies to something like Hide in Shadows. Unless, as a GM, we are willing to say no one but a thief (or halfling) can hide in shadows, then calling for a non-thief/halfling character to roll a DEX check allows them to be better at that particular dodge than classes that are supposed to be very good at it. (Forget the argument that thief's get better at it over time; a thief has to get to 6th level before getting a better than 35% chance to hide in shadows. And we would presume that a thief's DEX is far better than a 7 in virtually every instance.) Across the board, this method makes ability scores matter more than class abilities, and generally speaking a character would be better off rolling d20 ≤ Ability than rolling a class skill. 

Note that an ability check vs. an ability score of 17 (85%) is better than a 5-in-6 chance. Compare that to the Open Doors 5-in-6 for a STR 18. In other words, an ability check vs. a 17 STR is superior to an Open Doors check using an 18 STR. 

Yeah ... d20 ability checks are flawed. Sadly they are intuitive and therefor attractive to many GMs.

3d6 (or more) ≤ Ability (rows 4-6, shaded)

This is a clever way of doing things. It creates a very different curve. Note however that where this roll becomes superior to the 2-in-6 baseline is still pretty low. At 3d6 an ability of 9 is better. At 4d6, a 12. At 5d6, a 16. (These are signaled by red text.) Therefore, even if you ratchet up the difficulty with more dice, someone with a reasonably high ability will beat out a character with a prescribed talent. Often this means that the character themselves would be better going with an ability check. That is, a thief with a 16 DEX would be better off rolling 5d6 vs. ability than trying to roll a single d6 with a 2-in-6 chance. Again, this makes ability scores matter more than class abilities, which to me is a fatal flaw. On top of that, the GM has to make choices about when to call for 3d6, 4d6, or 5d6 rolls. 

d100 ≤ Ability (rows 7)

This method dramatically reduces the chance of success for "unskilled" rolls. It's a nice conservative way of doing things. It's also kind of obvious; your ability score is your % chance to succeed. Note that it's pretty rough if you are calling for rolls for things like running across a felled-tree bridge without slipping. In other words, only call for rolls when things are very risky. But overall I like this method because it makes rolling vs. ability score almost universally worse than rolling a class-based ability.

Dataset 2: showing methods on the x-in-6 or vs. 6 scale

Simple x-in-6 Check (row 1)

This method is probably the simplest and best. Most of the time the chance should be 1-in-6 (below the baseline) unless the character should, in your estimation, be naturally good at the task (perhaps considering a high ability score as a reason to increase the chance) or if the roll is for more of a die-of-fate kind of thing (like the prescribed method for triggering traps, p. 109 of the OSE Classic tome).

The Need a 6 Method (rows 4-7, shaded)

A final method is to have players roll d6s based on their ability modifier. If the ability ≤ 0, roll 2d6 and take the worst result. Otherwise roll a number of dice equal to the modifier (1d6 for +1, 2d6 for +2, 3d6 for +3). Success happens on a 6. The odds play out that only a +3 is better than a class ability.

d6 Modified by Ability ≥ x-in-6 (8 and below)

I have come to the conclusion that calling for a d6 roll modified by ability is also a bad idea. It feels intuitive to me, but it may force you to set various difficulties (sometimes over 6) and is still superior to the baseline 2-in-6 chance. It's just wonky. The less said about it the better, but I included it because I have done it. It seemed like a good idea at the table, but it wasn't. If you are going to do it, the TN should always be 7 or higher so it takes a +2 to be as good as a class ability, but this means people with no bonus automatically fail.


I think setting a simple x-in-6 chance is best. Personally, I would make this a roll-high thing, and pick a target of 4, 5, or 6 (easy, average, hard). I wouldn't let players roll for things that are exclusive class abilities. If I did let them roll for something like that (i.e. Hide in Shadows or Move Silently) that the difficulty would be 6, but it would require favorable circumstances for the character to even get the chance.