Dealing with Thief Skills

A common, agreed-upon tenant of OSR gameplay is the inclusion, encouragement, and necessity of player skill: that is, your dice rolls and your stat blocks will be less important than how you, as a player, set those character elements to use. Commonly taken to mean that a player should act out a conversation or barter rather than roll a reaction or Charisma check, or a player should describe spreading fine chalk powder along the floor seams rather than relying on your X-in-6 chance to find the secret door - Supplement #1 and the introduction of thieves flies entirely in the face of this principle: producing a series of actions that a Thief can attempt that - regardless of the player - has a chance of success based on the level of the Thief character.

Herein presented is a stolen idea - or, an idea formed from participating hedge-wise in a conversation between some unnamed OSR enthusiasts: so, not stolen, per se, but transcribed and translated from that conversation - to reconcile the Thief class abilities with player skill: improving the chances of success proportionally as the Thief levels but putting more agency into the hands of the player: literally, with playing cards.

Dealing Thief Skills

In lieu of rolling a percentile, when a Thief attempts to use a skill, the referee should produce a standard deck of playing cards and deals two cards to the player. The value of these cards is as follows:

• A numbered card bestows a number of points equal to its number.
• An Ace counts as 1 point.
• A Jack counts for 11 points; a Queen, 12 points.
• A King allows the player one of two choices: they may keep the King, which counts for 13 points - or they may request two additional cards: choosing one to keep and add to their pool. A second King drawn using this method should generally be treated as a 13 rather than produce a recursive extra draw - but this is at the discretion of the referee.
• Optionally, for simplicity, a referee may simply treat a King as a 13.
• Likewise optionally, if the cards dealt form a flush - that is, are all of the same suit - the result, success or failure, may be considered a critical result; elaborated on further below.
  

Alternatively, in the absence of playing cards (or in the preference for traditional resolution), each card drawn can be represented by rolling d12s.

• The rolled result, per "card", is the point result.
  
• A natural roll of 12 may be kept for 12 or traded for the player's choice between two rolls - similar to a King. Again, a referee may opt to treat 12s simply as 12s in the interest of simplicity.
  
• Up to four duplicates of a single roll result are allowed, re-rolling if the same single result is turned up a fifth time.
• To determine the optional critical (flush) - again, extrapolated on further below - 1d4 can be rolled for each die - presuming no duplicates in results - and if all 1d4 rolls come up the same, a critical result is achieved.

-- Edit --

How does this differ from the standard method? As written - with a target range to roll - it doesn't! Against a static target, it's simply another way to hit a predetermined probability!

It was my intent that a bust should indicate a failure with no retry until a level was gained, as normal, but other failures would allow a re-try: costing time. Do you chance it and try for a hit or risk a wandering monster by starting from scratch?

Apologies for the failure in editing - and thank you, community, for your continued feedback!

Note, the probability of achieving a critical and the nature of the skill element are slightly different between these methods: as it's more possible to get a "King" result with the dice than with the cards, but the King, itself, is more valuable when using the card-driven option.

Then, similar to a game of twenty-one, the player may request a buy (or hit), wherein another card is dealt to them (or they roll another 1d12), summing the result with their total. In order to succeed, the player's point total cannot exceed 24 points (double 12s on 2d12) and must exceed a target number, based on the skill in use and the Thief's level:

Level	Open Locks	Remove Traps	Pick Pockets*	Move Silently	Climb Sheer Surfaces	Hide in Shadows	Hear Noise
1	22	23	21	21	8	23	19
2	21	22	21	21	7	22	19
3	21	21	20	20	7	21	16
4	20	21	19	19	7	21	16
5	19	20	18	18	7	20	16
6	17	18	17	17	6	19	16
7	15	16	15	15	6	17	13
8	13	14	13	13	5	15	13
9	10	12	10	10	5	13	13
10	7	9	7	7	4	10	13
11	5	7	5	5	4	7	9
12	4	5	2	4	4	7	9
13	4	4	2	3	3	5	9
14	3	3	2	3	3	3	9

* When picking pockets, for every other level the target is above 5th, reduce the effective level of the Thief by 1 when determining target number: thus, a 7th level Thief picking the pocket of a 10th level Fighter would have a target number of 19.

Optionally, a character whose cards form a flush - that is, they are all of the same suit - a critical result may occur. On a critical failure, perhaps the tools have broken, the Thief has fallen badly, or perhaps the Thief has made a great amount of noise in the process - on a critical success, perhaps they have achieved a two-in-one, finding a trap and disarming it in one, climbed twice as far as expected, or perhaps finished early: producing a turn's worth of skill use in a round instead.

Again, I do not claim to have originated this idea - instead to be extrapolating on it as based on theoretical discussion - and as such I have not gamed with this rule yet. I think it might be fun - the use of playing cards is reminiscent of and thematically appropriate to first edition Deadlands, a tome lovingly collecting dust in my embarrassingly modest RPG bookshelf - the cards being something that would fit in to a weird-west theme very easily: the dice method, on the other hand, being a fit for any environment, their core mechanism being consistent with fantasy adventure play.

Also interesting to note - unlike other house rules I've posted here, this one is not designed to smooth play nor to speed it. It intends to effectively replace the Thief skill roll with a mini-game: as such, if your tolerance for adding complexity is low, this rule does add more rolls to the process, increasing the complexity of the over-all experience.

Regarding d12s

The math for the d12 probabilities is somewhat easy. The average roll on 2d12 - your initial "hand" - is going to be a 13, with a distribution as follows:

So it's possible to - on the initial hand, succeed on the nose (a proverbial blackjack), though fairly unlikely - 1 in 144, or a ~0.69% chance. Winning against a static dealer's hand on the initial hand, as you might expect, becomes statistically more likely as the target number shrinks:

Target No.	Chance to Hit	Target No.	Chance to Hit
2	100%	13	54.17%
3	99.31%	14	45.83%
4	97.92%	15	38.19%
5	95.83%	16	31.25%
6	93.06%	17	25%
7	89.58%	18	19.44%
8	85.42%	19	14.58%
9	80.56%	20	10.42%
10	75%	21	6.94%
11	68.75%	22	4.17%
12	61.81%	23	2.08%
		24	0.69%

A number exceeding 24 being a bust, or automatic failure, is impossible on 2d12: as such, the above values can about be taken at face value.

Recall, the above provides for the initial roll only: a target of 7 equates, on a single roll, to Climb Sheer Surfaces at level 3. For higher target numbers, which the Thief will inevitably encounter at low levels, they will be assumed to need to hit - or buy - that is, to roll again, bringing themselves closer to beating the number, but potentially going bust. Probabilities for various results, based on a third d12 entering into the mix, line up as follows:

Target No.	Chance to Hit	Target No.	Chance to Hit
3	100%	14	83.45%
4	99.94%	15	78.94%
5	99.77%	16	73.84%
6	99.42%	17	68.29%
7	98.84%	18	62.38%
8	97.97%	19	56.25%
9	96.76%	20	50%
10	95.14%	21	43.75%
11	93.06%	22	37.62%
12	90.45%	23	31.71%
13	87.27%	24	26.16%
	25 (bust)	21.06%

Note, the numbers above represent the chance, on the sum of 3d12, to meet or exceed the target number. Knowing that a 25 or above is a bust - one must consider that, for a target of 20, say: the odds are 50/50 that a 20 will be met or beat on 3d12 - but the odds are also just over one in five that the result will be over 24 - causing an automatic failure. Regardless, the target numbers provided in the rule's skill table are based on this one-hit (3d12) spread.

Why? Because:

1. The odds of going bust against 24 on 4d12 are in the vicinity of 40% - its average result being 26.
2. For target numbers below 10 - thus, target numbers for higher level thieves - the odds of hitting a success range on 3d12 are actually worse than on 2d12. The player will choose the course of action befitting the character's level and circumstances.
   
3. These numbers do not take into consideration the King effect - that is, roll-two-keep-one when a 12 is rolled.

Thus, 3d12 is a good baseline - knowing that player choice (and player analysis and talent at the mini-game) will offset the difference, definitively making the Thief more effective in their skills than on a flat percentile roll as a direct result.

Regarding the Deck

The math for using playing cards is more complex. You could treat it as having 13 possibilities, Ace through King, at which point the numbers would simply shift as for rolling 2d13 instead of 2d12. However, this is not entirely accurate - as dealing a card removes it from the deck: using a single deck of cards, this removal from the available remaining cards reduces the likelihood of hitting that card's value again from a chance of 4-in-52 (1-in-13 as above: or ~7.69%) to a chance of 3-in-52, or ~5.77%. 

The exact nature of this chance can be described using hypergeometric distribution: but for the sake of simplicity (and for some mercy on this poor soul's amateur statistician credentials), we'll consider the statistically similar probability curve on 2d13 and 3d13, respectively:

Are you better at statistics than I am? Prove it - hit me up!

Interestingly, this process produces an low but not insignificant chance to go bust on the opening hand when using a target of 24. This is part of the origin of the King effect: two Queens constituting the effective blackjack, two Kings would then require the player to draw again for at least one of them, avoiding the opening bust. Alternatively, a referee might opt - when not using the King effect - to simply set the target at 26: but that would then require recalculating the Thief numbers in the matrix in the interest of maintaining semblance of parity to B/X's Thief progression.

To Conclude

Having done the math, and having simulated the experience rolling some dice myself, I really do like the d12 method. I may have to implement it in a game and see how it flows, to see if it suits other folks as well as it suits me. Still going to keep the title of the post about playing cards though. Otherwise, it would ruin the pun. In either case, I hope the idea of Thief skills as cards suits your game, too. 

Delve on, readers!

Public domain artwork downloaded from OldBookIllustrations.com. Attributions in alt text.