D&D General D&D isn't a simulation game, so what is???

[Thomas Shey]

Would you call The One Ring simulationist?

I'd have to go back and look, but I seem to recall at least some simulationist elements; its predominantly a dramatist game IMO because of the impact of the character's potential flaws in overall game play arc but I want to say the travel rules seem somewhat simulationist.
(Though that gets into some of the issues with genre emulation there, since the source work has some pretty strong stylization).

 

[Thomas Shey]

For instance, everything in RM is based around d% rolls, -10 penalties and the like, and looking up charts. I don't think the inhabitants of GH can work this out. As I posted upthread, it is a type of breaking of the 4th wall.

I think it depends on what you mean by "work this out". I mean, honestly, you can probably establish a lot of that in the real world, and as percentages at that; that's part of what some sports statistics are in the end of the day. Are they going to think in terms of "modifiers"? Probably not, because modifiers are in the end a simplified representation of a much more grainy and varied process (and some of the die rolls themselves include some invisible modifiers simply because they're below the level of representation).

Putting to one side its enlightenment anachronism, one can imagine a character in the WoGH doing research to establish how things work in the world of GH. Imagining that does not depend upon using any particular RPG system.

Some RPG systems purport to model or represent - via their PC build, or action resolution, or setting-and-framing rules - some of the more salient processes and principles that the imagined research discovers. Rolemaster is an example.

That doesn't mean that the research will discover those RPG rules! I never have, but could, use RQ to play a game set in GH. That would be a different set of rules from RM, but that wouldn't change what an imagined in-world researcher would learn by researching the WoG.

Only if you matched the numbers up closely. Which would be ideal, but one of the things different games do is often assume somewhat different, sometimes strongly so, degrees of competence (or perhaps, what competence means in expression). Even different editions of RQ do this.

 

[pemerton]

Are they going to think in terms of "modifiers"? Probably not

That's sufficient to prove my point, I think. Because RM players, applying the rules of RM, think in terms of modifiers all the time.

So an imagined character who is not thinking in terms of modifiers has not inferred the rules of RM. Which as it should be.

More generally, it seems to me a total non-sequitur to infer from the fact that system X models world W, that inhabitants within W would, by research, infer X. All we can be confident of is that they would infer P, where P is the process (or set/system of processes) that occur in W, and that X models in some or other fashion.

It seems a sufficient proof of the previous paragraph that we might use a dice-based RPG in a simulationist way to play a game set in a world where dice have not been invented.

EDIT:
An even clearer proof is this:

No one can understand the rules of Rolemaster without at least grasping the notion of double-digit addition. But an inhabitant of the WoG might be innumerate and yet grasp most of what is true about the world that the RM rules set out to model - how fighting works, how running and jumping work, how talking to people works, how picking locks works, etc.

It's true that RM also has rules for knowledge skills, but that's one of the weaker parts of the system, and so the innumerate inhabitant who fails to work out anything about how maths and mathematical knowledge works in GH does not have a very big gap in their knowledge of the processes and phenomena that RM is concerned to model.

 

[clearstream]

That's sufficient to prove my point, I think. Because RM players, applying the rules of RM, think in terms of modifiers all the time.

So an imagined character who is not thinking in terms of modifiers has not inferred the rules of RM. Which as it should be.

Again, with sufficent careful observation they can infer that certain circumstances or actions change the distribution of results of other actions: these are modifiers. Your explanation would amount to saying that modifiers are not simulationist, as they have no causal correlation with the cosmos.

More generally, it seems to me a total non-sequitur to infer from the fact that system X models world W, that inhabitants within W would, by research, infer X. All we can be confident of is that they would infer P, where P is the process (or set/system of processes) that occur in W, and that X models in some or other fashion.

It seems a sufficient proof of the previous paragraph that we might use a dice-based RPG in a simulationist way to play a game set in a world where dice have not been invented.

Their representation might differ, that's true: dice or not dice isn't at issue. Suppose that the inputs and results of a simulationist mechanic S correlates with a reference cosmos R. A modifier for S is a circumstance C. An imagined inhabitant of R can conduct the following experiment.

Perform the action that S models - say throwing darts at a board. Where darts land is correlated with the results of S. (Otherwise S is plainly not causal.) Introduce a wide variety of circumstances that can bear on throwing darts on a board. Note down over many throws the differences in distribution of darts given the each circumstance. Among the list of circumstances there will be a C with a value sizing its influence on the distribution. In that way, the imagined inhabitant knows C is a modifer in the process S, and has a gauge of the size of C.

EDIT:
An even clearer proof is this:

No one can understand the rules of Rolemaster without at least grasping the notion of double-digit addition. But an inhabitant of the WoG might be innumerate and yet grasp most of what is true about the world that the RM rules set out to model - how fighting works, how running and jumping work, how talking to people works, how picking locks works, etc.

I will simply emphasise here the exact and conscious wording of my proposed definition "can know".

 

[clearstream]

@pemerton I can try to be clearer on something you might feel I am saying, but that I am not saying.

We can imagine a hidden system that produces results. Say a calculating device. It spits out a series of numbers: 4, 2, 9, 5, 1, 1, etc. There could be workings within the calculating device that are fully obfuscated from an entity observing only the output numbers. An example would be a process that adds 1 and is immediately cancelled out by another process that deducts 1. An observer of the results cannot infer either of those processes. A reasonable doubt you can raise is - is the calculating device a black box or not, from the point of view of an observer of the results?

There is a term in the proposed definition that says that models and rules take inputs and produce results correlated with references. So an observer must be able to write a set of rules that pragmatically-speaking map to the calculating device. Rules that when followed predict the same results. The observer can say things like - "Well in this case, I'd predict a 1" and they can be right about that (given sufficient data and numeracy... they "can".)

They are able to talk about rules that map to the rules of the calculating device, even though they can never be sure that there are not "silent" rules, and they might do things like reframe a dice roll as a graph (supposing they are familiar with graphs, but not dice.) Game designers don't typically indulge in silent rules. Modifiers are not an ideal example for you, because if they can matter to the results, they are not obfuscated and can be inferred. We can for sure find rules that can't be inferred and those are predicted to be typically metagame rules, rather than simulationist (even if the game overall is one we agree is simulationist). The definition isn't dismissing the possibility of such rules, it is defining that they are not simulationist.

 

[Crimson Longinus]

My preferred level of simulationism is some sort of relatively rules light and low-granularity system but still with simulationish sensibilities, meaning that the rules and numbers on the character sheet represent actual in-universe capabilities of the character and the rules of the game mostly represent in-universe causal processes. Some massaging of the odds for better genre emulation or playability is fine, as long as it is not too blatant. White Wolf's Storyteller system games are probably closest to my preference. D&D is somewhat less simulationistic than my comfort zone, but I still find its direction far more tolerable than the other extreme end of the games which try to be hyper detailed in their simulation and end up being unwieldy and tedious. (Yes, I mean you Rolemaster!)

Also I feel a lot of people confuse the level of detail with the accuracy of the simulation, and I don't think this is true. For example in Rolemaster the weirdly specific charts and results in them try to create an illusion of being scientifically accurate, but in practice they're often just jarring as even that mind-numbing amount of charts cannot actually account every eventuality and the results might often feel out of place or just bizarre given the fictional context in the game. I definitely prefer games which give more vague and approximate outputs that can be described and interpreted to suit the given context as needed.

When running D&D I try to interpret things somewhat simulationisticly, and I think the system works wellish enough for that if you're not trying to be intentionally obtuse and insisting interpreting a character dying from a hit of a bite attack causing piercing damage as them dying of embarrassment! Granted, I use gritty rests and healing kit dependency optional rules, so that makes keeping the HP somewhat associated easier.

I also intentionally try to make the setting metaphysics to conform to the rules when it wouldn't be too obviously silly, so for example, in my world the Nine Circles of Magic* are a known metaphysical concept, so the spellcasting rules indeed are pretty simulationistic for the setting! So in that sense I don't think these things necessarily need to be in conflict, and good design is one that can satisfy many criteria simultaneously, and I feel Gygax's choice of using Vancian magic in the game is a good example of doing just that.

(* Well, the Eight Circles of Magic are, the ninth level spells are so rare, that the Ninth Circle is just a theoretical concept or something rumoured in legends.)

 

[pemerton]

You might be reading in something that I am not saying. Let's ask some questions

• Can they infer the contents of that table i.e. the possible outcomes as facts about their world?
• Can they infer given careful observation the rate of those outcomes?
• Do we anticipate that they can experience some of those outcomes, in their world? e.g. a shattered knee (RM)?

It's true that they wouldn't learn that there is a rule instructing a player to consult a table. What they would know however is that if this happens then at least these outcomes (those on the table) are possible. A player can imagine their character talking about that - "Oh yeah, bites from giant ants? Could break bones. Might destroy organs... but that's extremely unlikely."

[EDIT So where a putatively simulationist mechanic is perfectly simulationist, it is necessarily true that it maps to facts found in the reference world.

Again, with sufficent careful observation they can infer that certain circumstances or actions change the distribution of results of other actions: these are modifiers. Your explanation would amount to saying that modifiers are not simulationist, as they have no causal correlation with the cosmos.


Their representation might differ, that's true: dice or not dice isn't at issue. Suppose that the inputs and results of a simulationist mechanic S correlates with a reference cosmos R. A modifier for S is a circumstance C. An imagined inhabitant of R can conduct the following experiment.

Perform the action that S models - say throwing darts at a board. Where darts land is correlated with the results of S. (Otherwise S is plainly not causal.) Introduce a wide variety of circumstances that can bear on throwing darts on a board. Note down over many throws the differences in distribution of darts given the each circumstance. Among the list of circumstances there will be a C with a value sizing its influence on the distribution. In that way, the imagined inhabitant knows C is a modifer in the process S, and has a gauge of the size of C.

As far as I understand these posts, they are saying that if a system is perfectly simulationist, then its probability distribution of the range of outcomes relative to inputs is the same as in the imagined world.

The framing of that claim in terms of knowledge enjoyed by the inhabitants of the imagined world seems to add nothing to the claim. I don't think it adds any clarity, either, to refer to the rules of the RPG system: the claim is not about the rules, but about a probability distribution of outcomes relative to inputs.

The claim it self seems to be intended to be a tautological unpacking of the concept perfect simulation.

I'm not persuaded it's a tautology, because I'm not persuaded that the notion of causal process is fully analysable in terms of probability distributions: and simulationist mechanics, as I understand them, are primarily aimed at exhibiting the in-fiction causal process, so that what the players are working out at the table via their rules and procedures (roughly) correlates with things that are happening in the fiction. The aim of the mechanics is not to predict what would happen in the fiction (unlike, say, the simulation that an engineer sets up in a wind tunnel). For the same reason, I'm not persuaded of your standard of perfection.

But accepting your account of perfect simulation, I can accept the tautology, but as I said don't see the benefit of rendering it in epistemic terms, or by reference to the rules of a RPG.

 

[clearstream]

the claim is not about the rules, but about a probability distribution of outcomes relative to inputs.

The claim is that simulationist rules can be inferred by the imagined inhabitants through observation and experience. We can see that goes beyond probability distributions by reflecting on diverse rules, like those for movement.

Maybe I have it worded in an unhelpful way in my proposed definition. It could be better to say something like the imagined inhabitants can discern laws that correspond to the simulationist rules. In order to avoid suggesting that they can know (or even think of looking for) the literal rule or the player-side processes (like the one you drew attention to, of looking up a table).

[EDIT One motive for the wording lies in @Thomas Shey's thoughts relating to the displacement of metagaming. Characters can know the rules, albeit we are speaking in the voice of conversation around the table. From the internal perspective it's better to say they can know laws corresponding to the rules.]

 

[pemerton]

It could be better to say something like the imagined inhabitants can discern laws that correspond to the simulationist rules. In order to avoid suggesting that they can know (or even think of looking for) the literal rule or the player-side processes (like the one you drew attention to, of looking up a table).

This seems close to what I posted upthread:

Even where RPG rules do represent things that the world inhabitants can know - eg how a -10 penalty in RM represents a modest degree of pain or impairment - the inhabitants don't know the rules of the RPG that we are all sitting around playing in the real world! They only know the represented things.

Putting to one side its enlightenment anachronism, one can imagine a character in the WoGH doing research to establish how things work in the world of GH. Imagining that does not depend upon using any particular RPG system.

Some RPG systems purport to model or represent - via their PC build, or action resolution, or setting-and-framing rules - some of the more salient processes and principles that the imagined research discovers.

I still take the view that causal processes are more significant than laws.

The goal of simulationist mechanics is to represent, and correlate to, ingame causal process ("internal causation is king", "the imagined cosmos in action"). This is the fashion in which they are not metagame, as per @Thomas Shey's post. Eg in RM, the attack hits, it bruises and it cuts - and the process of resolution at the table reveals this to us, unfolding it before our eyes: the attack is rolled, after allowing for defences it is applied to the table, looking at the column appropriate to the armour which indicates some of the bruising plus the crit (if any). The roll on the crit table fills in details of where and how severely the attack landed.

Ideally, there would be a single roll of d10000 that resolved both the attack table and crit table aspect: splitting them up doesn't reveal anything about the fiction, but is simply a technique for making the game playable using familiar dice and manual lookup tables. At the table we know there is a crit table; in the fiction, people know that a certain spread of injury is likely from that degree of success in attacking a person wearing that sort of armour. We don't need to posit that the world inhabitants have a concept of an 'A' crit, a 'B' crit, etc; that would be a bit cheesy.

We also don't need to posit that no one every suffers an injury not accounted for on the crit tables. The crit tables can represent a causal process without representing every possible variation that process might manifest in the fiction. Of course they need to represent what is typical and expected; and, as an integral part of the resolution process, they must yield concrete information about what is happening in the fiction. Ie they must do more than simply confer and constrain authorial permissions; they must dictate.

I'm influenced in the above by the following remark made by Ron Edwards: "Hero Wars character creation, . . . isn't concerned with the implausibility of having a mastery-level in Greatsword be just as "likely" as having it in Farming; RuneQuest character creation . . . emphatically is." This rules out crit tables that produce results that are implausible relative to what is expected in the imagined world. It doesn't demand literally comprehensive crit tables. (This is somewhat similar, at a structural level, to the Classic Traveller PC gen system: the fact that no one ever changes jobs in that system doesn't mean that career change is, per se, impossible in the Traveller universe. Just that it's not typical, and hence that, by using it, we get a good representation of the process of pursuing a career in the world of Traveller.)

This is why stronger claims about laws and predictions and the like seem unnecessary to me.

 

[clearstream]

I still take the view that causal processes are more significant than laws.

Hmm... for me, as I'm thinking about the imagined inhabitants, laws is right. In the sense of laws of nature, laws of physics, and so on.

We also don't need to posit that no one every suffers an injury not accounted for on the crit tables. The crit tables can represent a causal process without representing every possible variation that process might manifest in the fiction. Of course they need to represent what is typical and expected; and, as an integral part of the resolution process, they must yield concrete information about what is happening in the fiction. Ie they must do more than simply confer and constrain authorial permissions; they must dictate.

I'm influenced in the above by the following remark made by Ron Edwards: "Hero Wars character creation, . . . isn't concerned with the implausibility of having a mastery-level in Greatsword be just as "likely" as having it in Farming; RuneQuest character creation . . . emphatically is." This rules out crit tables that produce results that are implausible relative to what is expected in the imagined world. It doesn't demand literally comprehensive crit tables. (This is somewhat similar, at a structural level, to the Classic Traveller PC gen system: the fact that no one ever changes jobs in that system doesn't mean that career change is, per se, impossible in the Traveller universe. Just that it's not typical, and hence that, by using it, we get a good representation of the process of pursuing a career in the world of Traveller.)

This is why stronger claims about laws and predictions and the like seem unnecessary to me.

For me, it's valuable to have in mind the perfect simulation, because the games we call simulationist are believed to be imperfect simulations or approximations (as you explain above). That forces reflection on the matter of degrees of granularity: all are imperfect, none have 1:1 fidelity to the reference cosmos*. [EDIT When it comes to what they dictate, none are comprehensive and all are representative.] We can only be placing games and their mechanics along a scale, where none are perfect. That doesn't make all games and mechanics necessarily simulationist, but it does mean we can't exclude on grounds of being approximate or representative.

Laws in turn provide me with a wonderful filter for simulationist mechanics. I can ask - could the imagined inhabitants of the cosmos have experiences, expectations and behaviours that are compatible with the mechanics? If not, then my supposedly simulationist mechanic deserves further scrutiny... perhaps rejection. You have other lenses and from what you write it sounds like you are satisfied with the job they are doing.

*I'm divided on using "cosmos" because when I discuss an individual mechanic as simulationist I'm often not discussing the whole cosmos, but only one aspect of it. I feel drawn to parsimonious language in this respect, hence "reference".