The role of fortune

[Joshua A.C. Newman]

Gnarly. Thanks, Ralph.

[talysman]

OK, so, great, what function does randomness serve in... • GURPS

First off, as an iterative artifact, GURPS does not use dice in a consistent way.  You can't simply say "GURPS uses dice to mirror the unpredictability of the real world" or whatever.  Dice have been added to the design of GURPS for a number of reasons through the years of its development, often for very different reasons.

no, see, you're still looking at it wrong. you're looking at all the specific ways dice show up in GURPS, but not abstracting them into general categories, then asking "if we removed dice at this point, what would GURPS lose?" this is the fundamental question that Vincent asked for each of the games he analyzed above.

so, what kinds of randomness show up in GURPS, and what do they do to play?

• you have the three-dice bell curve, used for task or conflict resolution and for reaction rolls;
• you have threshold rolls, which determine whether or not Enemies, Allies, Dependents, Fickleness, and the like occur;
• you have table lookups for random effects;
• you have damage rolls, which are subtracted from some gauge (like Hits.)

#2 and #3 serve the same purpose: introduce unexpected effects, which may be purely descriptive or may actually limit or expand tactical and strategic options in play. #4 essentially fulfills the same function, but also aide in determining when a conflict has been won.

#1 determines the success or failure of specific tasks in the game. the fact that GURPS chooses a bell curve is important, because the GURPS viewpoint is that System = natural laws and that effects should follow a bell curve distribution, which allows you to predict what skill level you will need to get a consistent level of effect in the game world, while simultaneously guaranteeing that for average difficulty tasks, higher and higher skill levels have diminished effect.

it's all about making the game world seem predictable so that you can make character decisions, check to see if those decisions succeed or fail, and base your next decision on that result. task resolution leads to damage rolls which in turn resolve conflicts, and other rolls simply mix up the playing field.

• if you lose randomness in task resolution, players will concentrate on making a few skills sure-fire and won't risk using other skills;
• if you lose randomness in damage results, conflict resolution becomes more predictable -- you will always know after the first exchange who will accrue damage faster;
• if you lose the other two kinds of randomness, all conflicts begin to look the same.

d20 works much the same way, except that the designers weren't as concerned with the pretense of simulating "reality" -- so you get linear rolls on one 20-sided die for task resolution, instead of a three-dice bell curve. since target numbers for increasingly difficult tasks increase in a linear manner, this means that effectiveness (skill+attribute checks) gets into some very high numbers, which supports the emphasis on endless levelling as the primary game goal.

you also have the design decision that chargen should have random elements, too. thus, player resources during chargen are somewhat variable, and character design becomes a function of luck plus strategy, instead of just strategy as in GURPS.

[Josh Roby]

no, see, you're still looking at it wrong.

Still?  I just posted into this thread. o.O

you're looking at all the specific ways dice show up in GURPS, but not abstracting them into general categories, then asking "if we removed dice at this point, what would GURPS lose?" this is the fundamental question that Vincent asked for each of the games he analyzed above.

I was kind of answering Joshua (the other one), but okay.

#2 and #3 serve the same purpose: introduce unexpected effects... #4 essentially fulfills the same function, but also aide in determining when a conflict has been won.

So basically:

Overall, I would say that GURPS uses dice to support the game part of roleplaying game -- it adds an element of chance that theoretically makes the experience at the table more enjoyable for the participants.

That the 3d6 basis of GURPS creates a bell curve is not what the randomness is used for but how the randomness is expressed.  You're talking about the constraints that the randomness operates within (you will not get a 20 on 3d6) and the effects that this creates in play, which is great and all, but it's the second question after "What is randomness used for in GURPS?" which is the question that I was answering.

That said, I don't find GURPS' bell curve to actually function as it's presumably intended to.  Given the nigh-certainty at the end of the bell curve, all of my GURPS experience and all other experiences I've heard of have characters with reliable skills at 14+ (91% success rate), which is a relatively simple matter to produce with the point-buy character generation.  Like Hero, GURPS character creation is usually an exercise in min-maxing, and there is no systemic encouragement to make other decisions.  While effective scores in the 10-14 range are far less predictable and could conceivably make for an interesting game full of unexpected results (and this is what sample characters usually are statted with), I don't see that ever actually happening.  GURPS resolution dice as actually played are rather deterministic, not at all unpredictable, and if anything I'd say that given their probabilities they merely give the illusion that the players have cleverly beat the scenario when in fact the numbers were on their side since they made their characters.  In other words, they cease to introduce unpredictable effects.

This in turn led to the creation and prominence of the tables and crit-fail rules (18 is always a crit-fail, regardless of skill) which inject that sort of unpredictability back into the game, at the cost of injecting only generic unpredictability (you drop your weapon! ... but I was punching him) instead of anything with actual color or significance to what's actually happening in play (the price at which your success comes to you).  It's a patch-fix that does not correct the underlying problem, just paints over it.

All of which leads you to say crazy things like "First off, as an iterative artifact, GURPS does not use dice in a consistent way."

[ewilen]

I think both Joshua and John are correct that GURPS uses dice for a variety of reasons. I'd state it a little differently, though. Fundamentally, they introduce risk and unpredictability into certain aspects of play, which emphasizes certain types of strategizing and discourages others. (I'm not sure I agree with John's description of the importance of the bell curve in GURPS, but that's a detail which isn't worth going into here.)

In many cases, the reason GURPS uses dice is due directly to its philosophical and stylistic heritage in wargaming, via its ancestor The Fantasy Trip. In wargames, diced resolution is typically used in an attempt to model events which would be outside the control or knowledge of the players-as-strategic-actors in the situation being simulated, within the level of abstraction used for both decisions and outcomes. The use of a random model for this purpose is partly philosophical and partly an abstraction in itself: when I say that the distribution of darts on a dart board "can be modeled" using a random function, that's not the same thing as saying that it's truly random in real life. While this may be a difficult concept to grasp, it's not much different from the fact that a model of the Earth of the might represent the globe as a sphere (which it isn't): the model still represents the factors of interest to the observer.

In other cases the simulative motive is minimal, but the philosophical perspective remains. E.g., there's no particular reason that a spell, once learned, can't be cast perfectly every time. Or that casting it successfully might somehow depend on what would be seen as "metaphysical" issues relative to the game world. (Such as, "How important is it to the plot?") However, the need to make a roll turns says that spellcasting is an activity similar to mundane physical tasks: subject to success or failure based on objective factors which aren't completely in our control.

I'm going to gloss over other instances, but I want to acknowledge that, yes, the rolls for enemies appearing and whatnot are based on a perspective that's rather different from using dice for task resolution. There's still the basic philosophical stance: an impersonal universe where things happen unpredictably and without pattern. However, the way that taking an enemy ties into character creation and point balancing means that the dice are being used to flavor a particular metagame priority ("character balance") and turn it into a sort of gamble. Without that, there's no reason the Enemy disadvantage couldn't be handled through a kind of exchange of tokens or expenditure of resources.

[talysman]

no, see, you're still looking at it wrong.

Still?  I just posted into this thread. o.O

ok, I'll admit that I was mistakenly using "you" as a sort of mass-noun. "all you guys". I didn't check to see if you, Joshua, had actually posted to the thread before; I just saw a repetition of the same idea: not looking at the function of randomness in a game, but instead looking at all the ways randomness is used.

how it's used doesn't really matter. nor does whether the game really succeed, or whether it's broken for specific game styles; those are other questions for other threads. and I'm sorry for bringing up the fact that GURPS uses a bell-curve, since it confused matters and distracted from the question again; I only meant to mention it to show what the mindset of Steve Jackson was when he designed the game, so that we could analyze what the function of randomness was in establishing that mindset.

the function of randomness in GURPS task and conflict resolution is to make resolution of Situation predictable but slightly variable. you even mention that, Joshua, when you note that players will tend to min-max when designing characters. why? because they can predict that a 14 skill will succeed most of the time. if they want to regularly perform a specific task that always has a -5 penality, they will set the skill at 19, so that the adjusted skill is 14.

the function of the other random rolls in GURPS is to create variation in the Situation, which in some cases also affects the resolution. the rationale behind this, again, is that it's more "realistic", but the function is just variation.

[lumpley]

What none of you've provided yet is the purely procedural function randomness serves in GURPS. You've listed a bunch of ways that randomness contributes to what happens in the game's fiction, but nothing about how randomness contributes to what the actual people actually do.

Personally, I'm not surprised. I blame GURPS, not your collective analysis of it. Probably randomness doesn't contribute to GURPS' pure procedure in any consistent or identifiable way, on account of how GURPS' pure procedure isn't consistent or identifiable.

-Vincent

[Emily Care]

The payoff motivates players to start Complications which is a desired behavior...initiating Complications is a profitable activity.  But the uncertainty means that the player's behavior around Complications is much different than if that payoff were a guarenteed, known, non randomized event. 

if you lose randomness in task resolution, players will concentrate on making a few skills sure-fire and won't risk using other skills;
if you lose randomness in damage results, conflict resolution becomes more predictable -- you will always know after the first exchange who will accrue damage faster;
if you lose the other two kinds of randomness, all conflicts begin to look the same.

In each case here, the difference between with & without randomness in each case is clear and it is all about player behaviour.  The die rolls modify how players are likely to make choices based on the resources available to them. 

[jmac]

Isn't it GM's (or whoever takes most of responsibility) decision to choose if randomness will be used or not, anyway?
Is it really still a question - "what randomness gives us" or "how to use it"?

imho, using or not using dice etc is more about responsibility and making decisions in real situations - more of a feeling and intuition then system or rules.

[talysman]

Isn't it GM's (or whoever takes most of responsibility) decision to choose if randomness will be used or not, anyway?
Is it really still a question - "what randomness gives us" or "how to use it"?

imho, using or not using dice etc is more about responsibility and making decisions in real situations - more of a feeling and intuition then system or rules.

in a lot of older games, yes.

but I think a lot of people here would agree that there's something not quite right about that.

oldschool RPGs didn't have a clear procedure of play. mostly, they just had a clear procedure for combat. the closest they come to a procedure for play in general is in dungeon-crawl-style; the dungeon is basically a flow chart that links together combat and problem-solving scenarios. play is driven by movement from room to room.

Vincent is right that randomness does not really drive GURPS in any way. GURPS is one of those transition games from oldschool to modern day; it rose out of Man-to-Man, a pure combat game that was an attempt to recapture the feel of Melee. so GURPS has that same feel of oldschool play, with its clear combat procedures but not much else... but it was created at a time when game designers were beginning to think RPGs should be about more than just dungeon crawling.

the problem was: they removed the dungeon as the procedure governing play at high-level, but didn't know what to put in its place. all the games in this period did, at first, was add a skill system to the combat system and say "go forth and adventure"... when that was discovered to be too vague, they started applying the dungeon flowchart to the concept of plot, and you wound up with GM-designed choose-your-own-adventure, which of course lead mainly to Illusionist play or outright railroading.

this is why you see "metaplot" arising in this timeperiod. it's a "dungeon" on the conceptual level.

[Joshua A.C. Newman]

Welcome to the Forge, jmac! Do you have a real name we can call you by?

Isn't it GM's (or whoever takes most of responsibility) decision to choose if randomness will be used or not, anyway?

Well, lots of games don't have a GM, often distributing the responsibilites of a GM across other players. My games, Under the Bed and Shock: Social Science Fiction do that, as does Polaris. In many other games, the GM's responsibility is prescribed, for instance to the construction of a situation (like Dogs in the Vineyard), or providing contextual opposition (The Mountain Witch and Prime Time Adventures).

In all these games, the rules are never arbitrated by the GM. For instance, in Dogs in the Vineyard, anyone can call for a conflict to get something they want: the GM is without recourse if the players want to throw down on something. The rules literally say, in the GMing section, "say yes or roll dice". That is, either everyone shares the vision of what's happening, so it happens; or players disagree about what should happen next; or, in most cases, the dice wound the characters, pour salt in them, and they come out changed in ways you wouldn't have expected. That's why Vincent says they drive escalation.

You'll note, though, there's no facility for saying "We're not gonna roll dice about this." If the GM doesn't want something to happen, she's got nothing to do but throw some NPCs in there to fight with the characters over something - and truthfully, in Dogs, that usually doesn't kill the Dogs, so they get stronger from it, in the unlikely event that the Dogs don't win the conflict.

]Is it really still a question - "what randomness gives us" or "how to use it"?

imho, using or not using dice etc is more about responsibility and making decisions in real situations - more of a feeling and intuition then system or rules.

How you use those dice — what they do — is built into the system, just like your relationships with the othere players is, just like the rule on page 17 about underwater combat is, just like everyone's agreement that they can't meet Abraham Lincoln in this story because it will invalidate the last story.

The dice can't make you take or shirk responsibility unless you've already agreed to abide by them. They're clattering pieces of plastic that give you numbers. Those numbers inform the people who are playing the game because they've all agreed to read the dice a certain way. Then they people talk about what changes in the Shared Imagined Space due to the dice, which makes the players want to do something else, which they do according to the rules, maybe winding up back at dice or whatever Cues you're using.

What specifically it is that dice do in a given game, that has to be consistent and it has to confirm what's supposed to happen in that specific game. The only reason the GM shouldn't want to roll the dice over something is if either a) the rules don't call for a roll there or b) the rules are broken.

Does this response address what you were saying?

Now, at the beginning of this thread, lo these long years ago, I think I was probably talking about fortune in a Conflict Resolution situation, not every situation, but I also think I was unintentinally limiting the scope of the conversation, and folks called me on it. Thanks, folks! So the question has become, "In what ways is randomness used in specific games?" .... and its corollary, "In what ways do different games solve similar issues with different mechanics?"

[Joshua A.C. Newman]

... the dungeon is basically a flow chart that links together combat and problem-solving scenarios. play is driven by movement from room to room.
...the problem was: they removed the dungeon as the procedure governing play at high-level, but didn't know what to put in its place. all the games in this period did, at first, was add a skill system to the combat system and say "go forth and adventure"... when that was discovered to be too vague, they started applying the dungeon flowchart to the concept of plot, and you wound up with GM-designed choose-your-own-adventure, which of course lead mainly to Illusionist play or outright railroading.

this is why you see "metaplot" arising in this timeperiod. it's a "dungeon" on the conceptual level.

Holy crap, John, that was the most lucid way anyone's ever put that. You get a big, shiny Massachusetts apple for that.

[ewilen]

What none of you've provided yet is the purely procedural function randomness serves in GURPS. You've listed a bunch of ways that randomness contributes to what happens in the game's fiction, but nothing about how randomness contributes to what the actual people actually do.
-Vincent

I could speak a little more coherently on Runequest or The Fantasy Trip, both because I've played more of them and because their use of dice is more restricted. But again I would point to the "wargaming roots" of both games and say that dice are simply there as a simulative tool (according to a certain philosophy of simulation) within an overall freeform procedure. (TFT had many solo modules whose procedure was quite strict and dungeon-like, but the overall campaign context in the GM guide was every bit as "free" as other 70's to early-80's games such as Traveller or Runequest.) So why use dice at those points? I think a good portion of it comes from the interest in simulating combat and conflict, wargame-style; the real innovation in these games was the decision not to use dice in other parts of the game, even to eschew clear-cut procedure in those areas.

So the best I can answer your requirement, Vincent, is to say that randomness in RQ or TFT allows the players to address combat and certain other conflicts (find traps, silversmith, etc.) in a certain simulative fashion, at the same time assuming that other elements are better handled through pure judgment or consensus. The latter might also be done in a different simulative fashion (e.g., the GM makes a judgment based on what he thinks is likely to happen) but could just as easily be done in a dramatic fashion (the GM decides based on whatever makes a good story). Or the game might be a choose-your-own solo adventure with the GM just playing the adversaries. This means the dice may actually be embedded in very different overall game procedures; in some of them (Deathtest), the dice loom quite large; in others, they're really a small part of the game, though still reflecting a certain philosophical outlook on how certain conflicts will be approached.

If that doesn't address the issues you're looking for, I'd respectfully like to turn the question around and ask what purpose randomness serves in your choice of old-school game assuming whatever functional application of the rules text you please.

[Valamir]

I think Elliot is right.  Randomness in GURPS isn't there to support play from a procedural sense.  Its there to support a specific design philosophy, one that stems directly from wargaming and was such an ingrained assumption of the hobby of wargaming that I doubt SJ actually felt the need to articulate the purpose of its use in GURPs.

Basically the philosophy goes like this:
1) Reality exists
2) The outcome of actions players take in a game should resemble the outcome those actions would have if players took them in reality
3) That outcome should occur as a result of the same inputs that exist in reality (i.e. they happened that way because the inputs led to that outcome, not because you skipped ahead and just chose to have it that way)
4) If every concievable input was known, could be measured, and its impact calculated by an algorithm then there would never be any reason for randomness.  That is afterall how reality works...all of the various inputs combine in deterministic fashion to produce an outcome which would be completely predictable and known if those inputs were measureable.
5) Since It is not possible (nor desireable from a time and resources contraint standpoint) to account for every input and combination of inputs directly, all of the unaccounted for inputs must be accounted for using probabilities of outcome.

All of the above is just basic rules of modeling, the most well known example perhaps being Monte Carlo simulations.

The game design philosophy at work here is that the rules of the game should create a model of the games reality and that means accounting for as many inputs as can be reasonably ascertained (why there are tables and tables of situational modifiers in GURPS and its ilk) and accounting for all other inputs (known, suspected, or a total mystery) by means of probability. 

The most common method of dealing with probabilities in a game environment is via dice mechanics.

Voila...GURPS must have randomized dice mechanics because the entire purpose of the design is to serve as a model for GURPS's reality and that's the way models are constructed.

The impact of these mechanics on play procedures are only ever considered in terms of handling time (aka the old playability vs. realism debate) and "breaking" the system due to a poorly constructed model.  Actually using mechanics for specific player behavioral purposes or to generate a specific play experience was not a part of the equation.

Now whether this is something to "blame" GURPS for or applaud it for depends entirely on your feelings of whether Monte Carlo style statistical models of reality are a valid purpose for a game design.

[Marco]

I think Elliot is right.  Randomness in GURPS isn't there to support play from a procedural sense.  Its there to support a specific design philosophy, one that stems directly from wargaming and was such an ingrained assumption of the hobby of wargaming that I doubt SJ actually felt the need to articulate the purpose of its use in GURPs.

While I think this is a very good point--and, in fact, the material that you wrote that follows is pretty reasonable, I have a comment on it:

I can tell you for certain that JAGS, which uses the same model as GURPS in terms of mechanics (design patterns, if you will) did have a strong component of non-reality based deisgn goals and, in fact, I made a number of non-realism-based game-design decisions that centered around levels of player empowerment and predictability rather than a monte carlo simulation of reality.

Specifically, while I was concerned about the game system producing results that caused a strong break in suspension of disbelief, I was far more concerned with how the player felt about being in control of the action (which is why I moved from 2d10 to 4d6-4 early on). I can't be sure Steve Jackson didn't have those same considerations on a different level (or maybe he just liked the way Hero played--which we did too, who can say).

Now whether this is something to "blame" GURPS for or applaud it for depends entirely on your feelings of whether Monte Carlo style statistical models of reality are a valid purpose for a game design.

I think that the gestalt of the game rules plus a real group of human players will have a significantly different result during play than an analysis of the rules as a pure monte carlo simulation of reality suggests. I suspect a lot of the playtesters understood this and it's implicit in the design of the game.

When someone is actually using the system in a real RPG situation the situation itself is not one of a million, billion simultaneous situations in some computer simulation engine--it's part of a speciifc, highly artificial situation in a specific game. It may be part of a challenge, part of a story, or part of a "virtuality simulation." The rules, as written will support each of those modes of play very well when used congruently with the player's intent.

GURPS does not implicitly produce results contra-to-story or contra-to-challenge because of its "simulatory nature." In fact, one can argue, for example, that it produces pro-story effects by legitimizing and objectifying imaginary events in a way that is intuitively grapsed by players of an immersionist bent when the situation has some internal disposition towards rising action to climax, denoument, and so on.

-Marco

[Valamir]

I can tell you for certain that JAGS, which uses the same model as GURPS in terms of mechanics (design patterns, if you will) did have a strong component of non-reality based deisgn goals and, in fact, I made a number of non-realism-based game-design decisions that centered around levels of player empowerment and predictability rather than a monte carlo simulation of reality.

Specifically, while I was concerned about the game system producing results that caused a strong break in suspension of disbelief, I was far more concerned with how the player felt about being in control of the action (which is why I moved from 2d10 to 4d6-4 early on).

I was very careful to specify "GURPS's reality" and "the game's reality" to circumvent such "not a simulation of reality" concerns.

While I'm certain that "how the player felt about being in control of the action" was a design goal of yours for JAGS I'm not sure I buy the "far more concerned with" part.  I can see how moving from a wide dispursion die roll to a tight dispursion die roll is a nod in that direction...but if you were really "far more concerned" with players perceptions of controling the action there are many many other non GURPS-esque mechanics that accomplish that goal far better.  Clearly the monte carlo simulation aspect of that style of rules was important enough to you to make you not want to move away from it entirely.

I mean otherwise...using a design structure whose best feature is being a simplified-for pen-and-paper monte carlo simulator for you game would have been a really poor design choice, and I definitely don't buy that for a second.


I'll also note that far from millions or billions most practical common applications of monte carlo sims only run about 1000 iterations...strip it down to the much more limited number of variables we mere humans can handle easily at the table and a couple of dozen combats (easily achieved over the course of a old school length campaign) will give you a pretty significant result.

I also have no doubt whatsoever that once you have real players playing the game that it easily drifts to a great variety of purposes (and I never said anything about it being contra-story or contra-challenge that's totally out of left field) but it seems clear to me that GURPS and most other games of that ilk were designed around the ideal of modeling 1:1 scale conflict...just like wargames were designed to model larger scale conflicts.


I can't be sure Steve Jackson didn't have those same considerations on a different level (or maybe he just liked the way Hero played--which we did too, who can say).

Now whether this is something to "blame" GURPS for or applaud it for depends entirely on your feelings of whether Monte Carlo style statistical models of reality are a valid purpose for a game design.

I think that the gestalt of the game rules plus a real group of human players will have a significantly different result during play than an analysis of the rules as a pure monte carlo simulation of reality suggests. I suspect a lot of the playtesters understood this and it's implicit in the design of the game.

When someone is actually using the system in a real RPG situation the situation itself is not one of a million, billion simultaneous situations in some computer simulation engine--it's part of a speciifc, highly artificial situation in a specific game. It may be part of a challenge, part of a story, or part of a "virtuality simulation." The rules, as written will support each of those modes of play very well when used congruently with the player's intent.

GURPS does not implicitly produce results contra-to-story or contra-to-challenge because of its "simulatory nature." In fact, one can argue, for example, that it produces pro-story effects by legitimizing and objectifying imaginary events in a way that is intuitively grapsed by players of an immersionist bent when the situation has some internal disposition towards rising action to climax, denoument, and so on.

-Marco