Types of Inference in TTRPGs

[Idraluna]

This is a topic I've been thinking about for a while -- I wrote a somewhat inarticulate blog post about it in 2024 -- but since touches on some topics from recent episodes, I thought it would be interesting to discuss here.

In the Mythic Bastionland episode, Sam and Misha disagreed about whether some of the random tables were useful, and it seems like the core of the disagreement was whether the MB table results were specific enough. (As I recall, the specific entry was just the word "betrayal", which Sam thought too abstract to easily integrate with an existing narrative and which Misha regarded as workable).

I think their disagreement hinges on their respective preferences for types of inference. As I understand it there are three types of inference:

• Deduction predicts specific details from a general hypothesis.
  
• Abduction develops a general hypothesis from specific details.
  
• Induction extrapolates specific details into more specific details.
  

The famous and annoying example is the observation that Sherlock Holmes is less a master of deduction than of abduction -- his skill is in developing a successful hypothesis from specific clues. Deduction is only applied at the final step of pinpointing the culprit after acquiring a broader understanding of motive and method.

For how the relates to rpgs, my thesis is roughly:

Ttrpgs require inference to integrate exogenous inputs with the shared fictional game state. Depending on both the nature of the exogenous input and the way the fiction is understood and narrated, this process of inference can be deduction, abduction, induction, or a combination thereof. Preferences for one type of inference over another can illuminate some disagreements about play culture and scenario design.


Some examples of how inference is used in ttrpgs:

Referee improvisation in _Classic Traveller_: the referee rolls up a patron on a newly-visited planet. They have the following information: a small world with a tainted atmosphere and population of 10,000 people living under a corporate government, whereon a doctor patron has a job involving a playboy patron that includes an encounter with 11 thugs.

• The referee uses abduction to integrate these details into a story that involves the players (the playboy is a corporate nepo-baby put in charge of pollution control, his negligence is making people sick so the doctor has hired the players to intimidate him into doing his job, but the players will have to deal with the playboy's 11 lowlife boyfriends)
  

• The referee uses deduction to derive personal details from the general prompts "doctor" and "playboy", likewise for the planet ("tainted atmosphere" becomes "sky thick with clouds of iridescent industrial discharge")
  

Prepping a dungeon game scenario:

• JG's Wilderness Hexplore generates "Whickham, a castle, crumbling, manned by a garrison of 180 shaky mercenaries. Lead by kobold illusionist, level 3. (Neutral unless the leader's species suggests otherwise.) Walls are additionally defended by aerial creatures. Outside the walls, a stone bench." -- a slew of details with which to perform abduction and deduction.
  

• Luke Gearing's Wolves Hexfill table generates "Weird - Boon/ally" -- deduction is required to come up with specific details that fit the prompt.
  

Resolving a player action:

• A PBTA hack with a "partial success" result requires the MC or player to use deduction to narrate a specific type of partial success that fits with the existing fiction.
  

• Skills in Classic Traveller often have very specific results -- a roll of 8+ at gambling yields so many credits -- these introduce specific details into the ongoing narrative that feed into abduction and induction.
  

Investigation scenarios:

• In Brindlewood Bay (disclaimer: haven't played it, relying on second-hand understanding) players use deduction to fill in the details of a mystery from narrative beats.
  

• In a more OSR-inflected investigation scenario, players have to use abduction to piece together an understanding of a scenario based on specific details that are written into the scenario's key.
  

I have lots of other thoughts but two main implications are as follows:

First, I'm not well-versed in storygames but it seems like they often intervene in the fiction by supplying generalities, and that players derive enjoyment by "decorating" an exogenous narrative structure with specific invented details. (This might explain some of the backlash to Rules Elide, insofar as storygamers consider it "play" to backfill details which have been elided by game mechanics). Players who favor the OSR tend to prefer the process of developing a holistic world-picture from discrete, concrete details. I think this should at least be considered along side other play culture bifurcations like narrative agency/control.

Second, there are scenario design implications. Touching on another episode, one of the strengths of _Wolves_ is that deductive inference was used to create it, but it invites its players and referees to reconstruct that using abduction. Luke clearly had a preconceived idea of the Atlantean-Lemurian hyperwar from which he derived his hex descriptions, but by not providing a "here's what's going on" blurb he forces referees and players to think critically and make inferences. I suspect that in abducing their own interpretations of the setting, players and refs end up much more emotionally invested in the campaign than if the juicy secrets were just handed to them.

The two (probably inferior) alternatives are the "generality-only" approach (if Luke were to describe the Atlantean-Lemurian hyperwar and then ask refs to write their own hexes) and the "meaningless detail" or "pareidolia" approach where you generate lots of arbitrary description without there being a(n intentional) broader structure for players to reconstruct.

Some questions for discussion:

1. For those with a background in game studies, has this been written about or discussed before? Are there other theoretical distinctions this can be mapped onto?
   
2. Why do people seem to prefer one type of inference over another?
   
3. Per the examples above, it seems like inference happens at a lot of different levels -- is it reasonable to conflate the inference done by a referee when prepping a scenario with the inference done by players when running it?
   
4. Are there counterexamples where OSR games rely on deduction, or where storygames rely on abduction?
   

[misha]

Oh, this is really good. I don't have anything interesting to say on 1 or 4, but. . .

For 3, I do think that abduction is a huge part of what I love in play, and something I very much design my dungeons to support. I guess I do that with deduction (okay, maybe prep itself isn't inferring itself, but hopefully you get what I mean)? It does feel a bit like whatever the GM is using to prep is the mirror of what the players will then get something from doing mid-game; when I make a deduction in prep it lets the players fruitfully make abductive inferences.

And I think for 2, I feel like deductive inferences are more creative and feel better in the short term--for situations where you're creating the fiction rather than predicting it you end with with a result that's much more satisfying, but also once you finish with that specific instance you're done with it. Abductive inferences feel less 'creative', but also feel to me like they have longer effects on play; as a player in a megadungeon, if I can deduce the general forces shaping the dungeon I become better able to survive it indefinitely.

[Idraluna]

Oh, this is really good. I don't have anything interesting to say on 1 or 4, but. . .

For 3, I do think that abduction is a huge part of what I love in play, and something I very much design my dungeons to support. I guess I do that with deduction (okay, maybe prep itself isn't inferring itself, but hopefully you get what I mean)? It does feel a bit like whatever the GM is using to prep is the mirror of what the players will then get something from doing mid-game; when I make a deduction in prep it lets the players fruitfully make abductive inferences.

And I think for 2, I feel like deductive inferences are more creative and feel better in the short term--for situations where you're creating the fiction rather than predicting it you end with with a result that's much more satisfying, but also once you finish with that specific instance you're done with it. Abductive inferences feel less 'creative', but also feel to me like they have longer effects on play; as a player in a megadungeon, if I can deduce the general forces shaping the dungeon I become better able to survive it indefinitely.

I definitely share your preference for abduction -- most of this came from me trying to articulate why I like certain games and why certain tools are more useful than others when I prep scenarios. 

I really like the short-term vs. long-term distinction. It comports with the tendency for storygames to favor one-shots and games with limited scope, vs. the sometimes-realized promise of OSR games lasting for years and years. And the megadungeon example is perfect -- using abduction to figure out underlying patterns in the map or encounter table or history is part of mastering the scenario.

[sam_the_bold]

1. For those with a background in game studies, has this been written about or discussed before? Are there other theoretical distinctions this can be mapped onto?
   
2. Why do people seem to prefer one type of inference over another?
   
3. Per the examples above, it seems like inference happens at a lot of different levels -- is it reasonable to conflate the inference done by a referee when prepping a scenario with the inference done by players when running it?
   
4. Are there counterexamples where OSR games rely on deduction, or where storygames rely on abduction?
   

In order:

1. I can't think of any obvious connections? I would be willing to bet there is discussion of inference in the literature surrounding adventure games—that is, point-and-clicks of the flavor made by Sierra On-Line or LucasArts—but I have not read nearly enough to say for sure. I did some very fast searches on gamestudies.org, Analog Game Studies, and the IJRP for these kinds of inferences and found not very much. (A little bit of chatter around ludology-vs-narratology and how each derives meaning, and some stuff in Analog Game Studies about social deduction.) But this specific kind of detail-as-worldbuilding I did not see anything with immediate resemblence. 
   
2. Big question. My sense, in general, is that deduction favors styles of tables and play where discovery is de-prioritized—that is, where there is not some overarching extant truth(s) that need to be understood to progress. Like a lot of adventure-centric playstyles hover around some sort of general loop of transforming uncertainty/lack-of-knowledge into understanding and mastery. You map the dungeon, observe and understand monster weaknesses, learn the secret lore that lets you hit the boss's weakspot. Most of these, to my read, involve one of either abduction or induction, and correspondingly start you as players from a place of relative weakness and uncertainty. On the other hand, in styles of play that revolve less around cooperative problem-solving and more on creating drama and narrative together, it's important that everybody has a relatively strong shared sense of the world, even when the smaller details vary. Like if we're playing a game of Monsterhearts or Stewpot (in the way generally described by their authors), it's important that everybody understands the basic reality of play, because that's what allows each of our characters to make big spicy dramatic plays, the stuff that kicks up the drama. Similarly, as you say, filling in the precise details—say, the contents of my teenage monster's bedroom—is often very fun and expressive for players, and so these sorts of rulesets tend to allow for a lot of variety. We might call those details "fluff" or "flavor." Abduction and induction are really important for playstyles that hinge on discovery and transforming uncertainty into mastery, but for styles that expect players to create their own drama, deduction from general understandings allows a lot of expressive detail-making play—and the specific of those details are unnecessary anyways, since it's not players are ever really uninformed. 
   
3. I don't... think... so? Seems like no, to me, at least most of the time? Inference done as part of prep is largely to supply concrete details, no? Like it's to flesh out the world, to add diegetic detail and material for play. And that is, one way or another, nearly always about making concrete details—like it's basically always deduction or induction, in this sense, even if it follows some abduction. I suppose in a more player-authorial context, one could imagine a similar beat, where players collaboratively flesh out the world based on existing details, sort of like the Unscene in The Between or the leading questions in Stonetop? Kinda? But that's unusual. The other unusual overlap I can think of is in the context of a very tightly-written but very mysterious adventure—like the Wolves hyperwar, or some of the secrets in A Perfect Wife—where there are conclusions the GM needs to abduce out of the text as they read, similar to how the players will abduce those conclusions as they play. "Book : GM :: GM : Players" at its most literal level, sorta. 
   
4. I think deduction is relatively common for GMs in OSRish games when it comes to supply specific details. Like if the book or your prep says that the room has high ceilings but doesn't specify an exact measurement and a player asks "Can my 50' rope reach the chandelier?", your answer as to Yes or No depends on a degree of deduction, extrapolating "high" into a specific number. I think the difference is that many more Trve Simvlationists (like yovrs trvly) view that kind of deduction as something negative, something to be avoided, whereas for many GMs and players it's just business as usual. NPC reactions are another very common and harder-to-solve one: I know an NPC is grumpy and prone to insults, so I can deduce that when a PC asks for help, that NPC is likely to call them names rather than helping, even when that's not specified in the text. For the reverse, abduction in trad/storygames/neotrad, I think it's actually very common, it's just called "implied worldbuilding" or "anti-canon." Like when I spin up a new character and one of my starting gear options is, like, "Badge of the Planetary Constabulary," that, as the classic line goes, suggests something new and compelling about the world. Now the Planetary Constabulary is something more material, something we can hang our game around, potentially, which it (presumably) wasn't before. This is common-ish in more OSR and related projects, too—the key difference, I think, is that in more OSR-leaning stuff all those little details in chargen usually (ideally?) point to things that already exist, whereas in storygamier projects they tend to exist as jumping-off points for the table to create its own lore. Sort of abduction-as-mystery-solving vs. abduction-as-worldbuilding-prompt, if that makes sense.