The Primacy of Dice

[Lee Short]

Introduction

The release of 1st edition AD&D was the coming-out event for the hobby of roleplaying games.  Thousands of people across the nation played the game -- including many future game designers.  The game's influence on these future designers was deep, and the influence of AD&D on their designs was lasting.  Many of those influences would be shaken off in later generations of games.  Because AD&D sits firmly in the Gamist-Simulationist corner, it is the Gamist and Simulationist game designers that have had the most difficulty shaking off its influence.  AD&D gave the Narrativist game designer no basis to build upon, so Narrativist designers were forced to build from scratch.  It is the purpose of this essay to examine one of the most enduring legacies of that game:  the focus of attention during game play on the dice.  

Dice Primary vs Skill Primary

In contrast to most modern games, AD&D came with a plethora of different mechanics.  It seemed that each skill had its own mechanic.  But most of these mechanics fit into a single category: based on the character's ability, the character has a target number of some kind.  In order to succeed, he must roll above or below that target number.  In terms of actual game flow, this generally amounts to first calculating the target number and then rolling the dice.  This emphasizes the dice in the resolution mechanism.  By this, I mean that the resolution mechanism _psychologically_ relies upon the dice as the arbiter of resolution.  It is the dice themselves that finally determine the result -- and that is how the players tend to perceive it.  This legacy of AD&D has been continued in nearly every roleplaying game.  The slur "roll-playing" has a certain kernel of truth to it:  when playing these games, the players often feel as if the dice themselves control their characters' destinies.   I call these resolution mechanisms "Dice Primary".  

Contrast this with the FUDGE model -- where the resolution mechanism places the character's skill (and gamemaster judgement) in the driving position, psychologically.  The dice roll is centered about 0, and added to the character’s skill.  To the players, the dice appear to represent those “random” factors which cause performance variation.  Importantly, the dice roll is seen as a modifier to the character’s skill, and this connection is quite transparent.  The modified skill level is used by the gamemaster to determine the adjudication.  This method tends to make the character’s skill the primary factor in the resolution, in a psychological sense.  Even if the resolution mechanism is mathematically the same as a Dice Primary method, a Skill Primary method highlights skill rather than dice.  

FUDGE's influence in this regard has been quite minimal with respect to Gamist or Simulationist designs, perhaps because it's primary appeal is to a different crowd.  As a dyed-in-the-wool Simulationist, I find this unfortunate – as I believe Dice Primary methods have significant disadvantages.  

Thinking About Target Numbers

The first advantage of Skill Primary systems is that they aid the gamemaster in constructing Target Numbers (TNs).  The difference is in how the two types of systems ask the gamemaster to think about TNs.  A Dice Primary game typically asks the gamemaster to place the task in one of several difficulty levels.  This difficulty level then determines the target number.  Unfortunately, at no time is the gamemaster encouraged to ask himself the question 'how skilled should a character be to have even odds of success and failure?' but instead is asked the question ‘how would you classify this task on the scale Easy, Moderate, Hard, etc. ?’  I believe that the former question is a more useful way to think about target numbers.  I believe that through intuitively relating the target number to character skill levels, the gamemaster is presented with a natural and compelling way to think about target numbers.  However, Dice Primary game designs discourage this way of thinking about target numbers.  

Dice and the Game World

The second, and greater, disadvantage of Dice Primary systems is that they disconnect the dice roll from game world events.  To begin, let us examine what those dice represent in a Simulationist game.  For the game to qualify as Simulationist, the dice must represent something inside the game world.  Otherwise they should not be used at all.  Commonly, what the dice represent in the game world are the numerous micro-scaled factors which effect task performance.  Informally termed “luck”, these factors are too small to have an effect individually, yet have a macro scale effect in toto.   For a Simulationist game, the game mechanics should make it apparent what the dice represent: the whole point of such a game is that the adjudication is driven solely by events in the game world.  It should be obvious to the players that this is the case.  

I submit that Dice Primary methods obscure this connection, in two ways.  First, the attention that they place on the dice itself obscures the connection.  Secondly, they encourage game designers to use dice distributions with no connection to human task performance data.  The obvious flaws in these distributions in turn encourage the players to believe that the dice do not represent an in-game factor.  It’s a vicious circle, and it begins with the game designer failing to ask himself the questions ‘is the variation inherent in my dice method scaled appropriately to my skill level numbers, given what we know about human task performance?'  and  'does the distribution I am using match the distribution of human task performance?' and 'does the distribution I am using handle boundary conditions in a particularly unrealistic manner?'  

At least for Simulationist games, these would seem prima facie to be important questions.  But very few game designers seem to have given any of them more than a cursory glance.  Perhaps Simulationist game designers do indeed ask themselves these questions, only to determine that the problems are too hard to solve and/or that the traditional answers are good ones.  The most well-known innovation in dice mechanics is dice pools; other innovations such as FUDGE dice and “die minus die” mechanics have lagged in popularity.  Dice pools are more of a Gamist design than a Simulationist one, and this is apparent in the games that use them.  Their distribution changes shape and relative variance with the size of the pool.  They generally have boundary condition problems.  They also make the determination of sensible Target Numbers a task only for the mathematically talented.  Their large handling time tends to focus the game on the counting of successes – ie, on the dice.

Let us revisit the three questions above about the distribution of human task performance.   These are routinely ignored by game designers.  As a result, games routinely possess ridiculous models of human task performance.  These models are not only absurd to a mathematically sophisticated critic --- but they are so out of balance that any intelligent observer will find that their intuition is often at odds with the results of the game's performance model.  Commonly, this is because the variation in the dice method is much too high, and it is apparent that the dice roll is weighted more strongly than should be the case.  This reinforces the impression that events are resolved based on "dice, modified by skill" rather than "skill, modified by ‘luck’."  Another common problem is that boundary conditions result in hacked rules (if using 2d6, 2 always fails, 12 always succeeds, etc.) to avoid automatic successes or failures.  The hacks themselves always result in idiosyncracies -- such as additional penalties/bonuses having no effect on the chance of success.  These idiosyncracies encourage “playing the system” and strengthen the impression that the resolution’s basis is “dice, modified by skill”.  

From a Simulationist perspective, I have to decry these effects.  They are unsuitable for a pure Simulationist game, where the focus of the game should be on in-world events.  They do often suit the Gamist side of the Gamist-Simulationist corner where these mechanics originated.  

Looking Forward

All of this raises the question ‘what can we do to alleviate these problems?’  What properties in a dice mechanic will improve the situation?

First, the mechanic should have a very low handling time.  A mechanic which has long handling time will increase the attention given to the dice, subjectively increasing their importance in the resolution system.  Low handling time will keep the game’s focus on events in the game world.  

Second, the system should be a smoothly opened-end bell curve with no boundary conditions.  This is the type of curve which human task performance shows.  Matching real-world performance data will make the game’s resolution mechanism more believable.  

Third, the dice mechanic should be seen as a modification to character skill, which represents the effects of “luck.”  This is well-served by a distribution centered about zero, which sometimes is a benefit and sometimes is a penalty. The dice mechanic should also be scaled so that usually skill dominates over luck.  

Finally, luck should be more important in the short term than in the long term.  This means that long-term tasks should have a smaller luck factor than short-term tasks.  To do this, we must be able to scale our mechanic’s degree of variation.  

Two mechanics meet these criteria: FUDGE-like dice and “die minus die” mechanics.  

FUDGE-like dice can be made from dice other than d6’s, to meet the needs of games with smaller grain size.  For example, 12-sided dice with 6 blank sides and one side labeled each of: +1, -1, +2, -2, +3, -3 will create a smooth bell curve if used appropriately.  For example:  roll 3 of these dice, and roll one more die if all dice are of the same mathematical sign.  Keep rolling an extra die so long as the same mathematical sign occurs.  This method can be scaled to more or less variation by rolling more or fewer dice.  

“Die minus die” mechanics work by rolling two dice of the same size, and subtracting one die from the other.  These can be made to open-end fairly smoothly by awarding an extra die when a maximum value is rolled.  The variation can be controlled by varying the size of the dice rolled.  This method produces a curve that is less bell-shaped than that of most FUDGE-like dice.  This can be improved by rolling a neutral third die, which replaces the lesser of the other two dice if the neutral die’s value is greater.  

For games with a pure Simulationist ethic, these are excellent choices.  I would like to see more games use them, and any other mechanics which may meet the above criteria.  I have been using the first of these methods with satisfaction for years now.  The method has been successful in meeting my goals; I had one ex-player comment that he didn’t like my game system because “the dice don’t count for enough.”  It was music to my ears.

[Ron Edwards]

H'm,

Interesting, Lee. I think - at first glance - that one angle that seems to be missed is the role of dice as they relate to players, rather than to characters' abilities ("skills" as you call them).

If a game's system operates such that the dice apply mainly to the players' privileges (e.g. narration rights, character effectiveness, and more), then dice-rolling becomes exceptionally powerful for Narrativist purposes. Many such games are Dice Primary in your terminology.

At its most radical, you get games like The Pool, in which the Pool owned by each player only exists at the metagame level, and placing more or less dice from one's Pool into one's roll (increasing the chances of success) means literally nothing regarding what the character is using, doing, feeling, or anything else.

Closer to the traditional model, when you apply this system to "events in the game world" in a character-ability way, you get what I call Fortune in the Middle, which is exemplified in games like Hero Wars and InSpectres. (These games also include a certain degree of Skill-ness in your terminology, but not to the extent of say, GURPS or Fudge.)

Other games to check out in this regard include Otherkind, Dust Devils (OK, cards not dice, but big whoop), and Universalis.

Best,
Ron

[Lee Short]

Thanks for the pointers, Ron.  

I'd intended the essay to really only apply to Sim games, and evidently didn't make this clear enough.  I should probably have been more clear about that in the introduction.

Lee

[Ron Edwards]

Hi Lee,

Ah! Simulationist games, then? H'm, this will be an interesting discussion.

Dice in Sim games ... OK, historically, they've mainly been all about the "game-world physics" in the broadest sense, specifically reinforcing the real-world observation that doing Thing X does not always result in Thing Y, but usually clusters about Thing Y in some way from event to event.

Let's consider more limited aspects of play than "task resolution" or even "to-hit" rolls, which would be pretty broad. I'm thinking ... the degree that armor protects its wearer. Nice focused topic.

1) In most games, armor protects its wearer to a very fixed extent. It often follows the model set by The Fantasy Trip, Tunnels & Trolls, and early BRP (specifically RuneQuest), in which damage is calculated as a value of some kind, and then armor subtracts a fixed amount from the damage.

For games that follow this model, it interests me that so few introduce variation into this effect - only Stormbringer, if I recall correctly, actually introduced a roll to diminish the damage; if you wore plate mail, you rolled d8 or maybe it was d12 to see how many points were deducted from the axe's rolled damage to your character's tender flesh. I suspect that in these games, the desire to see "variability" in action was hitting a limit of handling time, which in games like Stormbringer (a BRP variant) was already considerable.

In other words, quite a few Sim games diminish the Dice-Primacy of protective values.

The other most-common system, set by D&D, decreases the attacker's chance to do damage in the first place (prior to a "damage roll"); this is what THAC0 is all about. This method tends to increase search time rather than handling time, but it also incorporates the (possibly desirable) variability in automatically.

I suggest that such systems tend more toward the Gamist end of things, especially that brand of Gamism that encourages a high random factor during play. (Damn! Must check Ninja Burger when I dig it out of the moving boxes; that's a pure Gamist random-heavy romp.)

One problem with generalizing like that, though, if that versions of D&D represent a non-independent cluster of design, which cuts down significantly on the available number of instances to consider. I don't know how Rune handles armor; can anyone help with that - is that a randomized variable? As a side note, I do think that Stormbringer, mentioned above, was a very Gamist slant on BRP (Ken St. Andre was the author, a dedicated Gamist in design terms).

But that's just one topic out of many. Anyone else want to pick some aspect of events-during-play that can be broken down into its Dice-Primacy status, for a given game or maybe how it's changed across games?

Best,
Ron

[Le Joueur]

Great article Lee!  I have a few problems with some of it, merely semantic quibbles for you to clear up.

Dice Primary vs Skill Primary

...based on the character's ability, the character has a target number of some kind.  In order to succeed, he must roll above or below that target number.  In terms of actual game flow, this generally amounts to first calculating the target number and then rolling the dice.  This emphasizes the dice in the resolution mechanism.

...I call these resolution mechanisms "Dice Primary".  

the resolution mechanism places the character's skill (and gamemaster judgment) in the driving position, ...the dice appear to represent those “random” factors which cause performance variation.  Importantly, the dice roll is seen as a modifier to the character’s skill, and this connection is quite transparent.  The modified skill level is used by the gamemaster to determine the adjudication.  This method tends to make the character’s skill the primary factor in the resolution,

...a Skill Primary method highlights skill rather than dice.

I'm not sure, but no matter how many times I read through this, the mechanics sound exactly the same.  One type, characterized by apparently bad design, puts too much attention on the dice, the other only appears to put skills first.  (And if you feel a need to contrast these to specifically, I'd be happy to analyze any die mechanic you care to describe to show that, mathematically, there is no difference.)

Really, are there clear differences between "Dice Primary" and "Skill Primary" from an objective standpoint or are we talking more like "any system is broken if players spend more time thinking about dice than 'in the world?'"  Many discussions have raged around here about 'low handling time' dice systems being as realistic or obscuring as you might want.

I guess what I'm saying is, are we talking about this difference:

Die Roll + Skill versus Skill + Die Roll?

That's what it sounds like.  The primacy of the dice is a highly subjective concept which I think comes more from how a game is written than from the underlying mechanics.

...Dice pools are more of a Gamist design than a Simulationist one, and this is apparent in the games that use them.  Their distribution changes shape and relative variance with the size of the pool.  They generally have boundary condition problems.  They also make the determination of sensible Target Numbers a task only for the mathematically talented.

Ron has made many good points that a game design that incorporates both Target Numbers and Variable-Size Die Pools is mitigating situations redundantly (I can't seem to find the link).  A game with parti-colored dice in pools of varying sizes would present results with extreme quickness.

First, the mechanic should have a very low handling time.  A mechanic which has long handling time will increase the attention given to the dice, subjectively increasing their importance in the resolution system.  Low handling time will keep the game’s focus on events in the game world.

You might look into Walt's discussion about die roll modifiers; sounds like it might be right up your alley.

Dice and the Game World

For the game to qualify as Simulationist, the dice must represent something inside the game world....

...Secondly, they encourage game designers to use dice distributions with no connection to human task performance data.  ... ‘Is the variation inherent in my dice method scaled appropriately to my skill level numbers, given what we know about human task performance?'  and  'does the distribution I am using match the distribution of human task performance?' and 'does the distribution I am using handle boundary conditions in a particularly unrealistic manner?'  

...Matching real-world performance data will make the game’s resolution mechanism more believable.  

Here's the main beef I have.  Since when is either real-world performance or believability important to role-playing games as a whole.  I'm more than willing to accept that realism may be important for a small slice of games, but as soon as you add in the common 'enhancements' (like magic, superpowers, future-tech, or whathaveyou), realism goes right out the window.  What's the point in having a realistic skill practiced by the guy standing next to the mage throwing a fireball?  I grant that many games could do with a good dose of verisimilitude, but realism?

Secondly, I think you might be confusing Simulationism with stressing the importance of simulation (an easy mistake considering the terminology).  Simulationism has emulation of reality and any kind of verisimilitude as secondary concerns at best.  Simulationism is about the exploration of one of the features Ron lists for games character, situation, rules, color, and setting.  That these are either consistent or realistic is of secondary importance.

Altogether, then, what are you asking for?  Better written mechanics that aren't complicated?  That doesn't solve the all problems you seem to have.  Mechanics that more accurately model reality?  That will necessarily be intrusive (making the dice, or system really, primary).  Or is it just a call for writing to make any system seem more about the players than the system?

Fang Langford

p. s. If you really want the article to be about Simulationist games, then shouldn't you have posted to the GNS forum?

[Ron Edwards]

Hey Fang,

Quick Moderator justification thing.

I'm happy with this thread in RPG Theory. It's utilizing GNS terms but not debating or clarifying them as a topic. The main issue is dice/randomizing, seems to me.

Also, what do you think if we specify to Simulationism, sub-set Exploration of System? That covers a huge range of existing RPGs and permits us to focus on things like "learning to pick locks," "taking damage from falling," "convincing the barmaid to perform oral sex on your character," and "hitting that guy with an axe."

Best,
Ron

[Valamir]

With regards to the first part of your post Fang, I think *appearances* are exactly what Lee was talking about.

The presentation of the "Dice central" mechanics draws the players attention to the dice as the primary factor to be concerned with.

The presentation of the "Skill central" mechanics draws the players attention to the characters abilities as the primary factor to be concerned with.

I don't think you'll find many actual significant mechanical difference (the illusion of FUDGE's adjective based system has been discussed many times).  But I do think that the PERCEPTION by the players during play can be very different.

I suspect this is mainly an issue when dealing with immersive play techniques where having mechanics that present skills first is less jarring than mechanics that continually draw attention to the dice.

[J B Bell]

I have to disagree about FUDGE putting primacy on skills.  Maybe it is in the text, but in actual play, the dice are terrifically variable; perhaps one of the most common topics on the mailing list, if not the most common, is how to mitigate  a really evil "whiff factor."

--JB

[Mike Holmes]

First Lee, good essay. That is, I love to see this sort of criticism (I do it myself a lot). But, I don't see this as nearly the issue you make it out to be. I'll address your points below, as they are numerous. I will say before hand that my attitude may be a bit caustic sounding, but it's not meant that way. Also, I do have the advantage of having been here a while, so I hope I don't come off as elitist. Please try to see it as an honest criticism of your (well-written) comments.

I will point out first (and I can't believe that Ron didn't mention this at all), for reference that use of dice and all randomizers falls under what is referred to as Fortune mechanics (from a Tweet essay, IIRC). But this does specifically address dice, and I will try to stay on target there.

AD&D gave the Narrativist game designer no basis to build upon, so Narrativist designers were forced to build from scratch.  

But oddly enough most still use dice. Ron has addressed this already, though, so I'll leave it there.

Contrast this with the FUDGE model -- where the resolution mechanism places the character's skill (and gamemaster judgment) in the driving position, psychologically.  The dice roll is centered about 0, and added to the character's skill.  To the players, the dice appear to represent those "random" factors which cause performance variation.  Importantly, the dice roll is seen as a modifier to the character's skill, and this connection is quite transparent.  The modified skill level is used by the gamemaster to determine the adjudication.  This method tends to make the character's skill the primary factor in the resolution, in a psychological sense.  Even if the resolution mechanism is mathematically the same as a Dice Primary method, a Skill Primary method highlights skill rather than dice.  

Interestingly, you say that the FUDGE method is better psychologically. Do you have any evidence for this? You admit that they are mathematically similar in some cases. So in those cases it's only psychology that gets in the way. But, other than personal bias (I could tell you that I don't feel that way, but that would just be my bias), do you have any evidence to back this up? I don't believe it's true, personally. If they are mathematically the same, then they seem exactly the same to me.

FUDGE's influence in this regard has been quite minimal with respect to Gamist or Simulationist designs, perhaps because it's primary appeal is to a different crowd.

I think that FUDGE has had some considerable impact. Considering its distribution scheme, the amount of talk it generates, and the relative lateness of its arrival, I think it's generated quite a lot of buzz. You aren't the first person to laud its design merits here. Do a search on the term Transparency to get an idea of how often this comes up.

In addition, I think that the Sim crowd is exactly the crowd that looks at FUDGE most often (it being highly Simulationist). Who else is looking at it more closely?

The first advantage of Skill Primary systems is that they aid the gamemaster in constructing Target Numbers (TNs).

I'll agree that there are mathematically better and worse ways to construct a system that informs players more simply about the odds. Again, this gets brought up in the Transparency debates as a form of transparency. Pretty much everyone agrees that the simpler this is, the better. That said, D20 is one of the simplest around. Is Fudge simpler? Perhaps, but the question is what other issues go with each. Not that I'm a big D20 fan, but looking at these things in a vacuum is denying that other factors can affect the usefulness of certain die mechanics.

I believe that through intuitively relating the target number to character skill levels, the gamemaster is presented with a natural and compelling way to think about target numbers.  However, Dice Primary game designs discourage this way of thinking about target numbers.  

In almost all of these systems there is a translation between words and numbers. In RM I pick Absurd, and that translates to a -130 or something. In FUDGE, you do the same math at some point. Just disguised by the terms. Again, this only seems to be a problem if people are really that distracted by numbers. Which I'm not at all convinced of.

To begin, let us examine what those dice represent in a Simulationist game.  For the game to qualify as Simulationist, the dice must represent something inside the game world.  Otherwise they should not be used at all.  

Why is this true? Why can't dice be the whim of the gods? Or the luck of the adventurous? Why must all Sim resolution systems be based on "reality" when, in fact they are usually set in unrealistic settings? I think this is very much a preference.

Commonly, what the dice represent in the game world are the numerous micro-scaled factors that effect task performance.  Informally termed "luck", these factors are too small to have an effect individually, yet have a macro scale effect in toto.

Yes, commonly. But this does not have to be the case. They can represent meta-game concerns. Yes, even in a Sim game. Or other things. This should not be an assumption.

For a Simulationist game, the game mechanics should make it apparent what the dice represent: the whole point of such a game is that the adjudication is driven solely by events in the game world.  It should be obvious to the players that this is the case.

This I agree with. But usually that's as easy as saying, "the dice represent the will of the Gods" or whatever they represent.

I submit that Dice Primary methods obscure this connection, in two ways.  First, the attention that they place on the dice itself obscures the connection.

Again, how? How does referring to a word on a sheet remind me that the roll is about something in-game and not that I'm playing a game. Why do numbers not do this? In any case I'm going to have numbers after I roll the dice. Aren't these going to detract?

Secondly, they encourage game designers to use dice distributions with no connection to human task performance data.  The obvious flaws in these distributions in turn encourage the players to believe that the dice do not represent an in-game factor.  It's a vicious circle, and it begins with the game designer failing to ask himself the questions 'is the variation inherent in my dice method scaled appropriately to my skill level numbers, given what we know about human task performance?'  and  'does the distribution I am using match the distribution of human task performance?' and 'does the distribution I am using handle boundary conditions in a particularly unrealistic manner?'  

This is a circular argument. If people use systems where the odds are worng, then they odds wont be right. I'm all for systems where the statistics of the outcome are taken into consideration, and due diligence is done to that. But what does that have to do with what you call Dice Primary games? Or is that part of he definition? That they have to be statistically flawed to begin with?

Moreover, the idea that FUDGE is in concert with reality in any way is ludicrous. The system is the way it is for the reasons you quote before, not because they have anything to do with "reality". And they certainly have nothing to do with setting expectations, either, as FUDGE is setting-less. So, Skill Primary systems can have just this problem as well.

At least for Simulationist games, these would seem prima facie to be important questions.  But very few game designers seem to have given any of them more than a cursory glance.  Perhaps Simulationist game designers do indeed ask themselves these questions, only to determine that the problems are too hard to solve and/or that the traditional answers are good ones.

I spend a lot of time here doing statistical analysis on Sim systems. Which "designers" are you referring to? Have some of them made these errors? Sure. But there are lots of Sim systems out there that have quite good distributions for delivering fun results. Some of them are mathematically just like FUDGE. Others even better. Given that they do employ different systems all the time (almost every Sim game has a new one), I think that they are at the very least thinking about it.

Dice pools are more of a Gamist design than a Simulationist one, and this is apparent in the games that use them.

What about dice pools mechanically make them Gamist? If there is a correlation with Gamist games it would be coincidence. Further, what mode a game supports best is very much a matter of conjecture. From my vantage I could argue that more Sim games use dice pools. In fact I believe they do. Given the usual increase in complexity that comes with dice pools, I'd even venture that the only reason for them is an attempt to get closer to in-game reality, and thus are a very product of Simulationism.

Their distribution changes shape and relative variance with the size of the pool.  They generally have boundary condition problems.

These are not true of all dice pools. In fact I have a system where it's not true at all, and can easily theorize more. Yes, bad systems are bad systems. Dice pools are not automatically bad systems.

They also make the determination of sensible Target Numbers a task only for the mathematically talented.  

Though they are still fairly intuitive, IMO. More dice = better. What more do you really need to know? In fact getting back to this subject, there are those who would argue that a player knowing the odds in advance is not "realistic" and that only a vague sense of it is helpful in achieving Immersion; a classically Sim goal.

Their large handling time tends to focus the game on the counting of successes – ie, on the dice.

True. Still this says nothing of all the multitudes of other systems that have arrived that do not use dice pools, but are still innovative.

From a Simulationist perspective, I have to decry these effects.  They are unsuitable for a pure Simulationist game, where the focus of the game should be on in-world events.  They do often suit the Gamist side of the Gamist-Simulationist corner where these mechanics originated.  

Interestingly, it could be said that all Random factors are more Gamist oriented. Why use dice at all? Why not just allow Skill to rule the day. Amber diceless does. As well as a few other systems. I mean isn't the perfect SIm system one then in which we know all the factors and can thereby just determine who does better in a conflict?

But I don't propose that seriously. I would agree with the usual counter-argument that one needs a level of abstraction, and at that point, this is where the dice are useful. At least in most settings and situations. But there is a question of where rolling makes sense in any game. Some Sim systems I've seen require one to roll when attempting a feat of strength, for example. But you rarely see a GM require a roll to pick something up. Instead of using Fortune in this case, one uses Karma (simple comparison).

The point is that any game is abstracting more or less of those elements that are referred to as luck. Even if we had a study that stated what the actual effect of Skill was, and could model it correctly (I do not pretend for a moment that any Skill l+ roll system can acurately portray reality with any fecundity), we'd see that it did not pertain to our setting for one of the many reasons I list above. Not least of which is that the setting and situation might not be 'realistic". Certainly the action produced by a Sim game like Feng Shui is not intended in any way to be "realistic" or even fair to the Mooks. Does that make it's resolution system bad?

First, the mechanic should have a very low handling time.  A mechanic which has long handling time will increase the attention given to the dice, subjectively increasing their importance in the resolution system.  Low handling time will keep the game's focus on events in the game world.  

Again, this seems to be a given. However, I don't see where FUDGE has less handling time than, say, d20.

Second, the system should be a smoothly opened-end bell curve with no boundary conditions.  This is the type of curve which human task performance shows.  Matching real-world performance data will make the game's resolution mechanism more believable.  

I agree that those are the trends, but the best a game system can hope to do is model. Given that, even if "realism" is the goal, one can certainly make concessions to these things where there is good reason. Otherwise, yes, I agree with all of this more or less. I would, however, point out that human effort is limited. A truly "realistic" system would have an upper bound to any task attempted by a human. And usually, there is a point at which the lower bound can be assumed to be the worst possible case. These things are easily model-able.

Third, the dice mechanic should be seen as a modification to character skill, which represents the effects of "luck."  This is well-served by a distribution centered about zero, which sometimes is a benefit and sometimes is a penalty.

Actually, I prefer systems where the dice give you something that is in the range provided by the ability. This is more "intuitive" IMO, than zero based.

Finally, luck should be more important in the short term than in the long term.  This means that long-term tasks should have a smaller luck factor than short-term tasks.  To do this, we must be able to scale our mechanic's degree of variation.

Or more simply just roll more to represent long tasks. The more you roll, the less luck matters.

FUDGE-like dice can be made from dice other than d6's, to meet the needs of games with smaller grain size.  For example, 12-sided dice with 6 blank sides and one side labeled each of: +1, -1, +2, -2, +3, -3 will create a smooth bell curve if used appropriately.  For example:  roll 3 of these dice, and roll one more die if all dice are of the same mathematical sign.  Keep rolling an extra die so long as the same mathematical sign occurs.  This method can be scaled to more or less variation by rolling more or fewer dice.

To use most other systems, just use the dice as they exist. Simpler.

"Die minus die" mechanics work by rolling two dice of the same size, and subtracting one die from the other.  These can be made to open-end fairly smoothly by awarding an extra die when a maximum value is rolled.

A favorite of mine. Made absolutely smooth if you simply add the extra rolled dice to the value of Max-1. Thus, if using d10s, when you roll a ten, roll again, and add to nine. Gives a perfectly smooth distribution.

The variation can be controlled by varying the size of the dice rolled.  This method produces a curve that is less bell-shaped than that of most FUDGE-like dice.  This can be improved by rolling a neutral third die, which replaces the lesser of the other two dice if the neutral die's value is greater.  

Or by rolling four dice, two up and two down.

I had one ex-player comment that he didn't like my game system because "the dice don't count for enough." It was music to my ears.

Odd statement. Isn't player satisfaction what we're looking for here?

Mike

[Clinton R. Nixon]

What's up with the total misrepresentation of Lee's post? I figured it out the first time, and what he said was so evident, I didn't post. I'm a bit confused at everyone else's interpretations of it.

Fudge (and similar systems) psychologically seem to put the skill first because the average attempt with a skill will result in the same quantity as the skill itself. In other words, if you have a Good skill, and use it, on average you will have a Good result. This is opposed to a game where you have a skill of +5 and on average have a result of +10, for example.

In addition, I think Lee did talk about the non-psychological factors of games like this. Fudge, for example, has a smaller distribution scale than, say, d20, where Fortune is much more important than skill for the first 5 or so levels.

What Lee appeared to say to me - and correct me if I'm wrong, Lee - is that in a system that attempts to appear to simulate realistic situations (note the difference between appearing to simulate realism and simulating realism - they're worlds apart), having skills both appear to be primary, and possibly being actually mathematically primary, is a good thing, and something that may not have been explored fully.

I've seen quite a few attacks like "but why is realism a good thing?" That's not the argument - games can focus on realism or competition or genre-emulation or story or whatever-the-fuck. What matters is "how do we achieve the goal we've set?" I think Lee may be on the right path here.

[Mike Holmes]

Lee said specifically things like "pure simulationist ethic", "I'd intended the essay to really only apply to Sim games", "From a Simulationist perspective", "let us examine what those dice represent in a Simulationist game".

I never once saw any limiting of that to a subset like realism. In fact, it seemed to imply that all Sim should be realistic. Which I disagree with. If your intention, Lee, was to speak to a smaller subset, then I apollogise. Consider all those remarks to just be an extended statement on my part on the variety of Sim, and having nothing to do with your point, then. I would warn you to watch your phraseology, however.

I still disagree on the psychology thing. I personally always find numbers to be more intuitive when describing anything with a magnitude than a word. I don't know how much better Good is than Fair, but I can tell you that 4 is twice as good as 2. Or give you the odds in a system. Put word in the way, and I have to translate before giving odds.

I may be strange in this, but it still seems to just be opinion. I'd like to see some justification for this statement about the psychology involved.

Mike

[Andrew Martin]

What properties in a dice mechanic will improve the situation?

First, the mechanic should have a very low handling time.  A mechanic which has long handling time will increase the attention given to the dice, subjectively increasing their importance in the resolution system.  Low handling time will keep the game's focus on events in the game world.  

Second, the system should be a smoothly opened-end bell curve with no boundary conditions.  This is the type of curve which human task performance shows.  Matching real-world performance data will make the game's resolution mechanism more believable.  

Third, the dice mechanic should be seen as a modification to character skill, which represents the effects of "luck."  This is well-served by a distribution centered about zero, which sometimes is a benefit and sometimes is a penalty. The dice mechanic should also be scaled so that usually skill dominates over luck.  

Finally, luck should be more important in the short term than in the long term.  This means that long-term tasks should have a smaller luck factor than short-term tasks.  To do this, we must be able to scale our mechanic's degree of variation.  

Two mechanics meet these criteria: FUDGE-like dice and "die minus die" mechanics.

There is another mechanic. My Ratio system:

> ...very low handling time...

Ratio uses a single D10 roll for about 90% of skill/attribute checks; two D10 rolls for about 9% of the time; three D10 rolls for about 0.9% of the time, four D10 in 0.09% of the time and so on.

> ...smoothly opened-end bell curve with no boundary conditions...

Ratio has infinite precision and never reaches 0% or 100%. It's bell curve exists in the skill/attribute percentile values.

> ...seen as a modification to character skill...

Ratio use Fortune in the Middle, so allowing players to add concessions or subtract complications from the D10 result so improving the PC's result.

> ...luck should be more important in the short term than in the long term.

Ratio uses percentiles, where the character and player know the exact failure rate, so that long term uses like crafting thousands of items it's easy to simply multiply failure rate by items produced to get the number of items that need rework. For example, a sentient PC machine with 99.96% skill in Detect Sniper Location will miss about 4 sniper locations in 10,000 attempts. Like Fudge, the percentile skill transfers across to the crafted item, and is the inherent quality of the item (how well it's made).

Of course, Ratio isn't better than Fudge in all areas. Ratio is only suitable for Modern, Sci-Fi and SF settings, where characters would logically know or can measure their own abilities with great precision. For other settings, Ratio introduces too much annoying precision into character descriptors. Fudge, with it's around 7 vague descriptors fits better in these settings.

[Ron Edwards]

Hello,

I shall do some more Moderating.

Mike, please try to minimize the line-by-line method of answering posts, for all the reasons you know. Once in a while is fine, but in a discussion like this one, it tends to breed like responses, and then everything goes kerflooey.

Also, one phrasing you used caught my eye ... you "warned" Lee regarding his terms. "Warning" is not something I'm going to permit from anyone but moderators, which in this forum - all of the non-publisher forums - are only Clinton and me, and it doesn't apply to terms debates anyway.

I'm pretty sure you were using "warning" in an intellectual sense only, and I appreciate your attention to the issue (I agree with the point you were making) ... but phrasing like "I think it's more correct to say" and so on is going to be more effective and less likely to be misunderstood. The potential for phrasing like "warning" to prompt a big bitch-session is way too high.

Best,
Ron

[Christoffer Lernö]

Lee, you're talking about something I myself have been spending a lot of time considering.

One of my own personal big beefs with many systems is excessive randomness. I've dealt with the subject elsewhere in detail. Anyway, as was argued there (not just by me), is that randomness is in many sim systems kept static despite more detail is known.

A better approach might be to reduce dice randomness as the givens are more well known and as the situation is less affected by randomness.

If I am a little better than Bob at playing cards, I might win over him more times than he wins over me. But the chances I have of winning a single game might be very random. On the other hand if we armwrestle it quickly becomes quite predictable who will win.

Or in other words, some situations are more predictable than others. This is almost always ignored.

Another thing is the effect of *when* to ask for die rolls. If I have to roll every 10 minutes to keep tracking or if I just make a single roll to see if I succeed or not makes a BIG difference in variation of outcome, and yet most games put in those rules arbitrarily with no thought of how they actually play out.

[Lee Short]

I had a long, somewhat different reply written yesterday afternoon -- but I resized my browser window to grab a quote, and it reloaded the page :-(.  

I must admit to having a certain amount of baggage with respect to the term Simulationist, as I was one of the regulars on rec.games.frp.advocacy while these terms were debated.  I think's that Ron's definitions are generally an improvement on the ones rgfa came to a consensus on.  

That said, I don't believe that this criticism is limited to 'realism', as defined in Ron's GNS essay.  Let's look at that definition:

The sub-set of Simulationism most fully developed during the 1980s was "realist" (a form of Situtation) and "genre-faithfulness" (System with strong and various other co-emphases). Some conventions of these approaches include identifying Fortune methods with the imaginary physics of the setting and a commitment to extensive search and handling times.  

First of all, I would identify Fortune methods with 'laws of entropy and uncertainty' or somesuch.  So I'm not sure if this is the same thing as what Ron means by 'physics' or not, but it may well be.   But my criticism is most certainly not restricted to games with extensive search and handling times -- as an example, off-the-shelf FUDGE very nearly meets my criteria, and off-the-shelf Tunnels and Trolls fails miserably (IIRC).

Its only real restriction is that it is restricted to games which feature skill performance of human or humanlike creatures.  Admittedly, this is a subset of Simulationist games and I should have considered that.  That is probably one effect of my past baggage: all the Simulationist games I was familiar with included such a mechanism.  

Let's examine an except about Simulationism from Ron's GNS essay:

The key to Simulationist play is that imagining the designated features is prioritized over any other aspect of role-playing, most especially over any metagame concerns. The name Simulationism refers to the priority placed on resolving the Explored feature(s) in in-game, internally causal terms.

Now, if the game has rules for human skill performance, then this resolution mechanism must be relevant in some manner to the Exploration being performed.  It may be a small part of that Exploration.  Even so, that part of the Exploration is still required to be resolved "in in-game, internally causal terms"  --  and my argument is that this requires a resolution mechanism with the properties I listed.  An in-game, internally causal representation of human or humanlike skill performance will match qualitatively with real human skill performance -- and, in a game designed in accord with my criteria, I believe that this will be apparent.  That last point is the crucial one.  It will not be apparent to every player, but it will be apparent to many players (or potential players).  

Further, a game designed with no regard for these or similar criteria will mismatch with real human skill performance -- and this will similarly be apparent, wrecking the in-game internal causality.  Once again, not everyone will notice -- but why not fix the skill mechanic and increase your game's appeal to more players?  

Admittedly, in some games Exploration of Character, the skill mechanic might be a small part of the game.   Perhaps even small enough so that "wrecking the in-game internal causality" is an overstatement.  But all that means is that the poor rules only wreck the in-game internal causality on a small scale.  Perhaps small enough to be barely noticed.  But isn't it still better to have good rule than a bad one?  

I guess my basic point here is that any game that includes rules for skill performance is, in Clinton's words, "attempting to appear to simulate realistic situations" to some degree or another.  Why not do a good job at it?