Simulationists, Black Boxes, and 4e

[Mustrum_Ridcully]

But what's wrong with just saying "Thog is what he is" and going with that, instead of always having to optimize him. (as you can probably tell, I'm not at all a fan of everything always being perfectly suited to its role and thusly optimized)The "warm fuzzy consistent feeling" is what world design is all about, and RPG design in my opinion should be about. When I sit down and design a world, I want to end up with something that the PCs were an organic part of before becoming PCs and could easily go back to being an organic part of afterwards; rather than have the PCs be special flowers...until they've earned it through adventuring.

I don't think I've put that very clearly. Every PC is or should be a commoner at heart, and every commoner should be able to join a party and adventure and (after some training) pick up the rudimentary basics of a class. 4e really skips over this step, and has the PCs start as pre-fab heroes without bothering to get them there, while ignoring whatever stages might lie between commoner and 1st-level PC. Which is fine, I suppose, if you want to be bad-ass from the get-go; but also woefully unrealistic. Me, I'd rather earn my badge of badassness in the trenches...or die trying; low-level PCs, after all, are there to be killed.

By the same token, what comes between the "minion" stage and the "skirmisher" stage for a Kobold? There's a *huge* difference between the two...far too big a gap to handwave away. Does a Kobold minion that survives a few battles suddenly wake up one morning with 27 hit points instead of 1? There's a bunch of gradations (or levels, counting up from a negative number to 0?) missing in there somewhere...

Lanefan

World-Building consistency only is required for things that are supposed to be measurable in the game world. Hit points are not something that should be measurable in the game world. Because if the were, this would quickly break any suspension of disbelief. Fireballs don't exist in the real world, but we can accept them in fantasy worlds because they are a conscious exception from the normal life, and most people can't use them. But hit points? No one in the real world has hit points. At least we can't measure them. So the creatures in the game world can't measure them, and they probably don't even exist there, either.

World-Building consistency generally concentrates on other stuff. Like ensuring that there isn't just a Metropolis in the middle of the desert, absend of any water sources or magical explanations for its existence. Or that if the King of Nowhere is battling against the King of Somewhere, they shouldn't suddenly ally the next day against the King of Anywhere and fight as one, barring a good explanation. Or if it is well known that the Archmage of Nowhere is a master diviner, he shouldn't suddenly be a master evoker the next day.
That's the kind of world building consistency that matters.

 

[mmadsen]

At the design level, I draw a distinction between "simulationist" (not the GNS usage) and "abstractionist" task resolution.

I strongly disagree with naming this dichotomy simulationist vs.abstractionist, because simulations can model "reality" at any level of abstraction, not just at an obsessive level of detail. In fact, you might just call your dichotomy detailed vs. abstract -- although most abstract systems aren't simply less detailed; they tend to be less concerned with being tied to "reality" at all.

 

[Nom]

But what's wrong with just saying "Thog is what he is" and going with that, instead of always having to optimize him. (as you can probably tell, I'm not at all a fan of everything always being perfectly suited to its role and thusly optimized)

The trick is what the abstractionist is optimising for. The optimisation is not for mechanical effectiveness; it's for gameplay effectiveness.

If I want a scene where the PCs get to face (and mow down) onrushing hordes, then I don't really want a detailed mechanical model for resolving the minutae of each individual participant. If I want a throwaway high-level wizard (life expectancy, 2 rounds), then I don't really want to make sure each of his 130 skill points are distributed correctly and that his entire spell book is coherent. If I want an ongoing series of one-on-one duels between a mighty wizard and a PC with lots of tactical options, then all this detail suddenly becomes important.

In a sense, some of this is just "levels of abstraction" (see below). But with the rider that different parties do not need to operate at the same level of abstraction. It's acceptable to run the PCs at their "normal" abstraction level while sliding a particular creature up and down between "detailed" and "mass combat". Because the mechanics don't actually tell the story in an abstractionist model, this is "no harm, no foul". To the simulationist, the mechanics are integral to the story-model, and thus this is seen as breaking the world.

I strongly disagree with naming this dichotomy simulationist vs.abstractionist, because simulations can model "reality" at any level of abstraction, not just at an obsessive level of detail. In fact, you might just call your dichotomy detailed vs. abstract -- although most abstract systems aren't simply less detailed; they tend to be less concerned with being tied to "reality" at all.

Which is exactly my point. It's not about "level of detail", but whether the resolution mechanics are even trying to model ("simulate") the same thing as the story. I considered describing "simulationism" as "concretism" or "coherence", but the former sounded silly and the latter is too imprecise (since an abstract system is supposed to be 'coherent' too, but using a much weaker mapping).

The critical difference is the extent to which we allow the player to stand between the mechanical abstraction and the story. In a "simulationist" or "concrete" model, the story and the mechanics must be synchronised before and after each resolution event. In an "abstractionist" model, you can synchronise whenever you choose and to the extent you choose, even if a large number of resolution events occur between the two.

Of course, nothing is pure. For example, a simulationist may accept that turn-based activity is an "unrealistic" game mechanic effect, so is willing to blur the story when it comes to who moves when. Similarly, few abstractionists would consider it reasonable to describe a character as having survived the combat when the mechanics clearly pronounce him dead.

 

[mmadsen]

In a "simulationist" or "concrete" model, the story and the mechanics must be synchronised before and after each resolution event. In an "abstractionist" model, you can synchronise whenever you choose and to the extent you choose, even if a large number of resolution events occur between the two.

I'm having some difficulty understanding what you're saying. Using your terminology, if I understand you correctly, a simulationist expects each roll of the dice to represent something real in the game world -- a hit or miss, say -- while an abstractionist does not -- a "to hit" roll may or may not refer to a hit, and a "damage" roll may or may not reflect any physical harm being done.

My gut instinct is to call the first good design and the second bad design -- not because the first is more detailed and the second less so, but because the first models something we're trying to model, and I don't know what the second one is even doing.

We could, after all, build up an equally abstract system -- that is, with little specific detail -- that still modeled something. We might, for instance, combine the separate to-hit and damage rolls into one attack roll, or we might even combine multiple such attack rolls into one combat roll. The system wouldn't tell us the details of how we got from here to there, but it would tell us the result: Conan defeats all the brigands.

 

[Kobu]

I didn't wade through multiple pages so I don't know if this was addressed, but the OP is not using the definitions properly.

There are four basic types of models: analog, descriptive (verbal), iconic, and mathematical. D&D uses these four types of models for different aspects of the game. What the OP is addressing are mathematical models--how the rules, charts, and stat blocks define how objects interact with each other.

Black-box vs. process-response is a distinction in (mostly) mathematical models showing how much we know or are interested in the inner workings of the model. With a black-box model, we would not know how any of the outputs would be determined. In D&D, we as DMs at least always know. The players may not know why they took 5 fire damage, but a DM can look at the mechanics and determine that the monster's free action allows it to radiate heat when bloodied.

The level of process complexity also has no bearing on making the models in D&D more accurate. Using ten different rolls and six charts to determine falling damage is just as inaccurate as using one roll if a person can always walk away from a 200' drop. The grapple example is poor because a more accurate simulation is not determined by level of abstraction. A single roll can determine that a better grappler wins.

What a simulationist is looking for is outputs that follow naturally from inputs. I expect a character that jumps in lava to die. If he does not, then the model is flawed. It really is as simple as that.

 

[Andor]

Maybe not in the same way, but there is an entire industry of scientific professional journals that would not exist if there was no need of interpretation of outputs (read data). To really understand science, one needs to realize that there is a level of interaction of reality that is so fundamental that is exists in a conceptual space that is outside of what humans can possibly experience. We can observe the outputs, create models of what we saw that describe what we saw, but we always have to keep in mind that the rule we created to describe the event we saw was not the real interaction between the objects in the event. It was just a description, and that is all we can ever get. We continuously revise these rules so that they better describe the totallity of all the events we have observed, but we will never get at the underlying interactions, and it is likely our rules will always have corner cases that do not fit in our existing rules.

You're over thinking this. There are elements of reality we can't perceive directly due to the limits of our senses. This does not make them some kind of Lovecraftian ultimate truth which would shatter our puny mortal minds if we could ever comprehend it.

A system of role playing that requires rules that are infallible in implementation and never create a "what the heck was that?" moment in the game would be so unweildy so as to be unplayable, if they were comprehensible in the first place. This is the reason for abstraction, fortune in the middle, and metagaming. They allow you to just hand wave the result to whatever you want so the game remains fun, comprehensible, and playable. Anyone who thinks that their simmulationist game will never have corner cases either uses liberal doses of abstraction, metagame constructs, does not understand the game, or isn't really playing it. Even in natural sciences this is true.

I keep seeing arguements like this made in favor of more abstract game systems and I've never for a single second understood them. When have you ever see someone decry a game system for not accurately allow me to track acetocholine levels in my characters brain? Or to instantly know the resonance frequency of his sword, or predict quantum tunneling events?

Never? Correct! No one cares about that level of precision. It's a degree of minutia that almost never impinges on play, and if it did, anyone would be happy talking it out with his GM. However knowing that a "hit" does indeed mean physical contact is a level of knowledge some of us want about our game worlds. The "You really missed but the dodge tired him" explanation makes little sense in any of a dozen corner cases, and there is the small matter of the fact that the basis for the D&D armour system is that most misses make physical contact but the armour protects the target from harm.

And I am not sure the natural science analogy is accurate for RPGs. RPGs are not for explaining data or predicting future events for which we have massive amounts of data to support. RPG rules are for organizing play and distributing power in a social context. It is actually much more like law than it is like science, even if rolling dice confuses the issue.

I don't think I buy that explanation for what RPG rules are. Very few RPGs actually concern themselves with the distribution of power outside of the the players get to narrate their characters, and the GM gets everything else. Even when it is mentioned it usually more along the lines of advice to GMs about how not to tick off players by telling them their character wet themselves in terror.

On the contrary being able to predict future events from very limited amounts of data is exactly what most RPG rules do. Will Joe Hero be able to jump the chasm? Can Darth Fido mind control the guards? Will the Count grant Vanessas plea? These are exactly the sorts of things RPG rules cover, becuase these are the events for which a neutral resolution mechanic (I.E. dice) will prevent conflict between the GM and players.

 

[PrecociousApprentice]

I think that I will address several posts at once here.

For those that have claimed that worldbuilding is nonsensical if any game constructs that describe the elements of the world are inconsistent in a one to one mapping fashion, with mutability of the results/game elements being a weakness of the process, I would point to the many millions of published fiction books out there that require no game constructs or concrete modeling at all to create fantastic and very detailed worlds. No one to one mapping of game concepts to world building required. These examples seem to be zero to one in their mapping of game to world. The world building is entirely fluff. No crunch required.

The point of the rules are not to create the world. Your imagination does that. The point of the rules are to organized play and distribute power within the playgroup, and to provide an agreed apon action resolution system framework so that the process can be a game. This doesn't require a one to one mapping of game construct to game world in exactly the same way that collectively writing a novel would not require this. That being said, most people prefer not to interact in a system of anarchy, so rules are important in creating fun even for the collective writing of novels.

I can understand that there are people that prefer that the world building logically emerge from the rules of the game in a one to one mapping fashion. This is one way. It inherently depends on the degree of infallibility of those rules to create worlds that are not jarring or bizare. There are many people who would likely self identify as sumulationists who would fall into this camp. I was once one of them. I was very frustrated for quite some time because there were many logical inconsistencies that would emerge in play. I eventually drifted away from gaming because the inconsistencies were too jarring for me and my group. I think that there will be a lot of frustration about 4e because of this type of problem. The rules when taken to relatively simple logical conclusions turn out some jarring results, and we haven't really even tried hard yet.

I would like people to understand that many people do not require that there is a one to one mapping of game construct to game world. These gamers have no problem letting the world and the game exist in parallel conceptional space, with the game helping to guide the creation of the world, but not expecting that the world be an emergent property that springs forth in a perfect state as a result of applying the rules as written.

There have always been, and will necessarily always be abstractions in the rules that will lay the groundwork for imperfections in the world if the world is expected to be a logical consequence of the rules. The human brain is the most amazing computing object known, but there are definite limits to it's capacity. Saying that we will never be able to create a model that we could use to predict outcomes of any complex system without logical inconsistencies does not imply any Cthulu mythos to be true. It admits that from our side of the equation, there are limits, and from the universe's side of the equation, illogical outcomes might be unavoidable. Advanced science is coming to grips with this. Gamers, working with less resources and smaller brains (myself included) will either need to come to grips with this or be doomed to frustration at some point.

4e seems to have embraced a design paradigm that sidesteps the problem at the cost of frustrating many gamers that don't embrace the paradigm. This is unfortunate. The problem for many remains as great as ever, WotC just ignored it. I also think that the problem is unsolvable, and I think that ultimately the paradigm that 4e sprang from will create more rewarding games. It seems to have created a more robust gamist platform, empowered narrativist play greatly, and sidestepped the problems of strict simulationist gaming, much to the chagrin of those players who prefer sim.

EDIT:

I don't think I buy that explanation for what RPG rules are. Very few RPGs actually concern themselves with the distribution of power outside of the the players get to narrate their characters, and the GM gets everything else. Even when it is mentioned it usually more along the lines of advice to GMs about how not to tick off players by telling them their character wet themselves in terror.

On the contrary being able to predict future events from very limited amounts of data is exactly what most RPG rules do. Will Joe Hero be able to jump the chasm? Can Darth Fido mind control the guards? Will the Count grant Vanessas plea? These are exactly the sorts of things RPG rules cover, becuase these are the events for which a neutral resolution mechanic (I.E. dice) will prevent conflict between the GM and players.

I think that I should address this. The rules do not describe actions. They give a framework for action resolution and a division of power, upon which the division of power allows those with current authority to describle the event in a wany that is enough in accord with the results of the action resolution to make the game fun for all. No one to one mapping of mechanic to game world result. The mechanic guides the description, but doesn't prescribe much at all except whether there was a success of failure. These can be described many ways. Frustration emerges when imperfect rules are not allowed a descriptive fudge factor. Recognizing a role that the rules fulfill that sidesteps the logical results of imperfect rules allows one to play with much less frustration.

 

[Doug McCrae]

I think more abstract rules, such as T&T, support simulationism better in that they produce fewer WTF moments. By virtue of telling the players less about what is going on, they break suspension of disbelief less.

Imo the best system for not breaking SoD is completely freeform - GM decides everything.

When it comes to superhero gaming I'm very much a simulationist. I love the crazy comic book universe but I don't think any rpg has come remotely close to mirroring it. You'd need rules for characters being more powerful in their own books, remembering and forgetting their powers as required by the plot and having a different personality depending on who the writer is this week. Nobody but me really seems to want this level of madness however. Strangely this kind of sim isn't consistent at all. Not when you're trying to simulate an inconsistent universe.

With D&D I feel the great strength is the core rules - class, level, hit points, etc - not the simulation of fiction or reality. They're not particularly G, N or S. Their major selling point is ease of use, their simplicity.

 

[PrecociousApprentice]

You have sumed things up nicely Doug, and have hit on why I either don't like GNS at the moment, or I completely don't understand it. The words simulation, narrative, and game do not necessarily intuitively imply what GNS says they are. Less ambiguous terms would help.

 

[Primal]

there are different kinds of internal consitancy... and i am missing a bit here and there:

Two comparable monsters (like kobolds), one wears armor and has higher dexterity than the one without armor: who should have higher AC?

Of course the one who does melee... but does it hurt if you give the melee one scale armor to reflect his AC? the result is the same (which is actually the important part of monster design) but you have some kind of consitancy...

And yet although that armor exists in the story and the mechanics, it is not available to PCs as "loot", because it is a given that PCs are not interested in monsters' equipment (not even if it was magical). Yet what if your players argue that they want to sell kobold armor in town? What if one of the monsters has a better armor (in mechanical terms) than one of the PCs, and the player would like his character to use it?

To me it seems that 4E mechanics and story exist on two "layers" of reality: PCs live on one, and the rest of the world live on the other. Whenever an encounter/challenge happens, those layers partially -- but never completely -- interact with each other. And maybe it's just me, but this feels... weird?