I had an idea roughly midway through my aborted sabbatical. Roughly speaking, it is this: the way games use dice fall into distinct and easily catagorized patterns. This is worth an in depth study, both as an intellectual study (for amusement) and on a more practical level as a means of identifing useful and useless mechanisms when discussing game design.

Of course, dice are not the only means of resolution/randomization in games, but they are the primary one.  

I apologize if this is a bit unfinished/rough as I did not have the time to actually hash it out when it was still fresh in my mind.

 First you have 'single die' resolution.  There are a few subcatagories here.

Vs.

Addative

Subtractive

Vs is exemplefied by things like D20 and percentile systems like BRP.  Essentially the dice are rolled and compared to a number. Very commonly there is little gradiation in level of success. You either succeed or you fail.  Levels of automatic or critial success and failure are common, but this still leaves a lot of leeway in measurement.  This means that a great deal of potential variable goes unused by the system, but in the hands of a creative GM this 'open territory' allows for a lot of flexibility. Close calls and easy shots can be based on how comfortable the margin of success or failure is.

Addative is when you roll a single die and add that to a number. The primary difference on the surface is that your 'difficulty' is no no longer bounded by the facets of the die. I'm thinking of things like Mechwarrior 3, SLA industries and Cyberpunk 2020, where the variable is actually smaller than the numbers added to it. One could suggest that there are actually two addative catagories: Greater and Lesser, describing the overall impact of the die on the task.  Cyberpunk is the exemplar here, where it is possible to have the addative be double (or more) than the variable. That is 20 plus a d10.  This reduces the importance of variables, making a 'skilled person' more important than a 'lucky person'.  Note that there is some cross over here.

Subtractive: Notional. I can not think of any system where the dice are subtracted from the fixed numbers as standard.  I hope to expand on this, or hear/think of an example to study.

The second major division is, easily enough, the pool system.  For the purposes of discussion, this does not include 'fixed pools' especially of small size (two dice pools, etc). This catagory is much wider than the single die catagory.
All pool types are typified with difficult 'scaling' of variables compared to the single pool method.

Single High pool: Classic Shadowrun, Heavy Gear(though more hybrid...). These are cases where you want to roll many dice with the goal of getting a single die as high as possible.   More dice grant greater opportunities to hit the high note, but imporatantly, even the smallest pool can acheive great, even overwhelming success. A single Die in Shadowrun can acheive, potentially, the same 'maximum' as a pool of ten or more dice can, it's just much less likely.

Multi High pool: This is Legend of the Five Rings/Seventh Sea catagory. Roll and keep.  Alternatively, Heavy Gear fits here, with the ability of multiple sixes to add to the overall success. Here the size of your pool can overwhelm even a lucky opponent, but good rolls are still a must.  Personally, I find this slightly inferior in terms of variableness compared to the Single-High pool.  In cases like L5R, the size of the pool is inferior to the number of Kept Dice in determining the overall level of ability.

Seperate Dice-Fixed Target: This is the 'hit counting' school. Currently the darlings of Burning Wheel/Empire and Shadowrun 4 use this method.  This method has an advantage over other pools in that the math to determine the 'rate of success' is actually quite easy. For example, in both Shadowrun4 and BW/E, a 5-6 on a six sided die is a success. So on average, every three dice thrown is a success.  If something requires three success to 'go' then you need an average pool of 9 to succeed as often as you fail.  On the other hand, unlike the other methods pool size is the only value of note in most cases, and I've noticed that they do not feel very... granular.  Since the high level of averaging, coupled with the fixed averages, you very rapidly get into a mindset of 'pass/fail' gaming when it comes to dice/target numbers.  While the actual dice throw may not reflect the averages, over time this makes for less tension in any given series of throws. It also means that players tend to either look for ways of lowering difficulties (if possible) or increasing dice to the exclusion of all else.  Again, this may just be a feeling coming from discussion.  Obviously, less easy averaging numbers may be a solution.

Many dice-variable target: White Wolf has expiremented with this at times. You want to count success, though due to the changing of the target numbers, often a single success is all that is needed. This gives the GM an unparralleled level of control over how success is measured, yet is the least useable in that regards.  Determining with any finesse the 'difficulty' of any die throw requires either a great deal of 'feeling' for the dice, guesswork, or fairly complex charts and math.  This works fairly well for games that are run more on instinct rather than planning, but can quickly fall appart as things tend to swing wildly from too easy to too hard in the hands of many GM's.  Of all the pool mechanisms this one is the most unworkable, I feel.

Reductive Pools: The only example that springs to mind here is Dying Earth, though I know it is not unique. This is a pool where the number of dice are merely 'available' not all thrown at once. This is a radical departure from normal pools. Here the amount of the pool represents your ability to keep struggling to succeed, often allowing you to force a 'lesser' opponent to conceed even if they initially won, simply because they can not hold on to their gains.  I find it reasonably elegant and playable, but suspect it is rare due to the difficulty in making the system work.  In theory, any pool based game could probably be turned into a reductive pool, provided there is a wide enough range, or narrow enough as the case may be, to make the game playable.  I believe it only really works with fixed target numbers, though the potential to try it with variable target numbers should not be discounted.

Bell Curved Dice:

This is essentially a variation of the Single Die mechanic using multiple die for the same effect, resulting in greater averages in tosses.  Rather that discuss each variation, they will be treated as a whole. Typically, GURPS is a good standard for this, roll 3d6 compare to skill.  In fact, this should go under the 'single die' catagory.  

The important difference is of course that the number of dice thrown decreases the variables of the dice, until with a high enough pool you get a statistically flat value.  This is 'serious' dice rolling, that is it removes a lot of the 'odd ball' results. the hero does not succeed easily one swing then miss wildly the next. The goal, presumbably, is consistancy.  In the interest of fairness, I should point out that I prefer this sort of dice mechanic overall. It is simple and consistent at the same time, others prefer the wild randomness of true single die systems.  

Significantly, bell curved systems work inversely to their single die counterparts. The larger the single die, the greater the variable, the greater the 'bell curve pool' the smaller the variable.


Comparative Dice: this is the Feng Shui, Qin:the Warring States, and presumably d4-d4 system. This is where two dice are rolled and one is subtracted from the other. This results in a smaller variable pool, with the added bonus of allowing for negative die results, depending upon the system.  In Feng Shui, certainly, the fixed value in any roll is the most important of all, and this will hold true, I suspect of all such mechanics. Qin, I belive, removes the potential for negatives by always subtracting the smaller die.   This, I believe, is a slight failing of the system as one can presume that the actual dice mechanic could then simply be replaced by a smaller variable die for a single die addative system.  I have noticed that some players have a difficulty grasping the subtraction mechanism at first, but the learning curve should not be viewed as a serious impedement as it is still very short.  


This is not exhaustive or refined by any means. If I have missed a fairly fundamental catagorization, by all means let me know.  Likewise, I have not yet begun to discuss Gimmicks in mechanics.    While I consider Gimmicks completely seperate from standard mechanics, I have noticed a few games that use gimmicks so extensively and fundamentally that they probably do belong in a catagory in and off them selves.

An example of a Gimmick is the Yin/Yang die in Qin, where the color of the larger or smaller die is important depending on the type of throw (red die high in combat means bonus damage). More commonly, Unknown Armies uses extensive dice gimmicks with their flips and other 'tricks'.  Other gimmicks are simply replicating an existing mechanic is a bizzare new form (for example a percentile system using  d6's and a base six mechanism. Still essentially a percentile throw...) or wild dice being added.

Also, you have hybrid systems to consider. I have noted that Heavy Gear seems hybrid to me, with the typically small dice pool, the use of a single, heavily modified die to determine success or failure compared to a target number etc...

Nice analysis, O Sparky One! Got some interesting groupings. I like that you group small die pools like GURPS and JAGS with single die rolls, because that's how it feels to me too. I think you've covered the basics really well. Is theis leading to something?

And real good to see your pointy ears here again!

-clash

I don't know that it's leading somewhere, but over the last five or six years of gaming I've increasingly become aware that every new game I buy is just some variation of a game I already have... at least as far as dice are concerned.

Ideally I'd like to make this a tool so that gamers and game designers could look at what the actual mechanics out there are like and find something that fits their style, and vice versa.  Likewise, we have an assumption that you could look at what is out there and compare something you already have made up and see what it actually feels like, and adjust it.

I'm reluctant about the Gimmick label, but to be honest, that's what they seem like to me. Gimmicks.  They spruce things up, but don't meaningfully add to the process. In fact, many may even make determining 'average difficulties' harder... which is equally a GM's issue as a Player issue.

It's something I originally came looking for on the Internet when I stumbled on all this other "theory" stuff...randomization mechanics.

I'd love to see a tool that said - if you take approach X, here's the odds, etc.

So if I use d20, I know it's a certain shape of random and has these effects.  If I use 2d10, that shape is more like this with these effects.  If I translate that to a pool of type A, it looks like this...and so on.

Lastly - how would you classify (if you even would) computer-based randomization?

Quote from: James J SkachIt's something I originally came looking for on the Internet when I stumbled on all this other "theory" stuff...randomization mechanics.

I'd love to see a tool that said - if you take approach X, here's the odds, etc.

So if I use d20, I know it's a certain shape of random and has these effects.  If I use 2d10, that shape is more like this with these effects.  If I translate that to a pool of type A, it looks like this...and so on.

Lastly - how would you classify (if you even would) computer-based randomization?

Most randomization programs are designed to emulate a specific dice mechanic in that regards. Straight randomization, without looking at replication of dice, would probably be the 'single die type', where you can control the size of the variable much more exactly than you can with dice.

By itself, it's just a gimmick.


As for theory: yeah, this is the sort of thing I would look for in theory too...

Quote from: SpikeI don't know that it's leading somewhere, but over the last five or six years of gaming I've increasingly become aware that every new game I buy is just some variation of a game I already have... at least as far as dice are concerned.

"You say that like it's a bad thing!" says the game designer who never did anything original in his life.

Quote from: SpikeIdeally I'd like to make this a tool so that gamers and game designers could look at what the actual mechanics out there are like and find something that fits their style, and vice versa.  Likewise, we have an assumption that you could look at what is out there and compare something you already have made up and see what it actually feels like, and adjust it.

That shounds really good!

Quote from: SpikeI'm reluctant about the Gimmick label, but to be honest, that's what they seem like to me. Gimmicks.  They spruce things up, but don't meaningfully add to the process. In fact, many may even make determining 'average difficulties' harder... which is equally a GM's issue as a Player issue.

Some people don't seem to care, especially the abundant math-phobes out there. I think designers should know, but it's not really necessary to know otherwise - though I much prefer mechanics I understand.

-clash

Quote from: flyingmiceI think designers should know, but it's not really necessary to know otherwise - though I much prefer mechanics I understand.

Like I said, understanding the ramifications was the reason I came looking in the first place.  And it's not just knowing the odds, but the patterns of the odds (Bell Curve, Log, etc.), the impacts to which they translate (more experienced characters will have a better chance than low level with luck, etc).

That would be golden. EDIT: An Encyclopedia of this, with each "type" as Spike provides discussed.

Quote from: SpikeMost randomization programs are designed to emulate a specific dice mechanic in that regards.

Agreed, though if I remember my geek Comp Sci side, the way randomization actually works, underneath the layer that represents dice, can vary.

Quote from: SpikeStraight randomization, without looking at replication of dice, would probably be the 'single die type', where you can control the size of the variable much more exactly than you can with dice.

By itself, it's just a gimmick.

Well, not really.  I mean, one of the things you can do with the computer is get beyond the straight-line randomization into things like logs and such - ya know, take the random number and then perform some function on it.  I don't know if that's a good thing or bad, but it opens up some other avenues I think - maybe allowing a simple mechanic to be scaled up to Superhero levels or something?

As I've said before, though, I think the best of all worlds are dice that tell the computer (through RF or IR) what was rolled.  Which brings us back to understanding how different mechanics affect goals.

I'm not sure how much I should dwell on/ignore the touchy feely aspect of dice.  You know, a big honkiing handful of dice is going to make some players happy because they can SEE how good or bad they are every time they get ready to roll. Others will naturally get annoyed at dealing with that many dice, and are happiest with one die, because it's all right there....

Which of course is only a sample of what I could go into with 'touchy-feely' dicing.

I think I missed a good one. I know I was thinking of it when I wrote the OP, but somehow I never got into it.

Multiple Variable Dice: This is your typical 'change dice to reflect ability' type, be it Deadlands, HPL, Earthdawn, Alternity or any other.  

On some levels this could be viewed as a variation on existing resolution mechanics. Often the dice rolled are singles, other times they are used for a shifting 'bell curve' single dice variation. In Deadlands, notably, they are used for a 'High Single Pool'.

However, due to the wide sweeping effects of this mechanic, I feel it is it's own unique catagory, with the capacity to draw elements from the other catagories.  Actually, I think it almost necessitates a 'meta' catagory, or perhaps seperate entries in the existing catagories.

One of the more interesting points of using this sort of dice mechanic is that while your 'lower end' remains more or less constant, changing to bigger dice invariably make your average and maximum rise.  

The exception to this rule is when using variable dice with the 'exploding' sub-mechanic. Smaller dice will explode more often than larger dice, skewing results at the bottom end without actually altering probabilities at the very high end.  For example a difficulty of 29 is virtually impossible with d4's, even factoring exploding dice. For a d12 it is highly improbable, yet for a d20 a single explosion of the dice is 5%, and acheiving a 9 on the second throw merely reduces your probability to roughly 3%...  Thus we can see that in most cases adding the exploding mechanism doesn't statistically alter the value of higher dice, it merely reduces the gaps between 'steps'.

Interesting stuff Spike. But whenever I see this kind of classification of dice mechanics, it always feels as if they don't really cover the elegant, but unique system used by Dogs in the Vineyard. Yes, it's not a widely-played game, but it's mechanics do make it an excellent "stress test" of this kind of classification

It seems to me to be a mix of "Multiple Variable Dice" (I have 3d6 Acuity, 2d4 in "I'm blind as a bat", and 1d10 in "I shoot real good") for scores. But they get rolled into a pool, and you make attacks (called "Raises") by taking any two dice from this pool and adding them together to generate a total (so I might take the d4 which rolled a "4" and the d6 which rolled a "3" from my pool to generate an "attack total" of 7). Which seems a bit "bell-curve-y". (This pool persists for the entire conflict, by the way).

And then the guy whom I attacking needs to beat my total by taking dice out of his pool that sum to my attack total or more (or just Give up and lose the overall conflict). The number of dice he uses to do this (which is called a "See") shows how well he blocked/dodged and how much potential damage ("Fallout") he takes from my Raise. That seems to me to be kinda "vs." with a bit of something else, as the number of dice used is very important.

And you've got the whole escalation thing going on, but that's kind of orthogonal to the raw dice mechanics. Escalating from "talking" to "fighting" (or whatever) usually lets you roll extra dice into your pool, but makes any Fallout you receive more serious, to put it simply.

Managing your dice pool is one of the best parts of the Dogs dice mechanics, IMO (I have a rolled 1,1,3,5,6,9 in my pool; he Raised me a 10. Do I See using the 1,1,3,5 and take some Fallout from it, but leaving my big dice to make a big counterattack with the 6 & 9 (=15), or do I take my big 9 and a 1 now to block his attack but be less effective when I Raise back at him (the biggest I could then make back in response is a 5+6=11)? Or do I Escalate and do something serious to him in return for being able to roll more dice into my pool, but that means Fallout has got a lot more deadly? Or do I Give up on this conflict and cut my losses now?) It doesn't really seem to fit into any of the categories happily.

I mean, I understand the desire to be the unique snowflake and all (mine goes to 11!), but...

OK, all snarkiness and kidding aside, why is that not "Multiple Variable Dice"?

It might be that other mechanics of the game make the inplementation feel different, but...

Just curious.

EDIT: And this very question is why an Encyclopedia Randomica is required...

Hey Spike!

Here's an interesting dice mechanic you haven't mentioned!



-clash

Quote from: James J SkachI mean, I understand the desire to be the unique snowflake and all (mine goes to 11!), but...

OK, all snarkiness and kidding aside, why is that not "Multiple Variable Dice"?

Fair points. To my mind, I can "feel" the commonalities between say Burning Wheel & nWoD (both "Separate Dice-Fixed Target", to my mind) or CP2020 & d20 (Both "Additive Vs.", pretty much). But I can't "feel" any similar relationship between DitV to Deadlands or Savage Worlds. I guess a mechanical relative could be Button Men, as no other RPG I know of has that "manage your rolled dice pool over several actions" kind of feel to it.

And it's not wanting to just be "unique snowflake"; it's just that the Dogs dice mechanic is the strangest I have ever come across

Quote from: James J SkachIt might be that other mechanics of the game make the inplementation feel different, but...

True; but where do dice mechanics stop and the rest of the game start? What I described in my previous post I would say are the dice mechanics for Dogs, but I can see cases to draw the line in other places.

Quote from: James J SkachEDIT: And this very question is why an Encyclopedia Randomica is required...

True enough.

Clash: Can't see the picture at work, sorry.

Warren: Actually, you do remind me of a rather rare mechanic, of which you add the 'variable dice' attribute to.  This is a type of Reductive Pool I think, at least from what you described. While it is significantly different from the Reductive pool in, say, Dying Earth in that you apparently control how many dice you throw, the pool is your limit of ability.

The rest of the mechanics (raising, stakes, etc) are beyond the scope of this discussion. However, at some point I hope to discuss things like Opposed roll mechanics and so on.   DitV, from everything I've heard, is one of those rare games where you only roll against another 'person'... which includes the GM.

Quote from: SpikeClash: Can't see the picture at work, sorry.

Rats! That ruins the fun!

What about ORE with it's yahtzee-like mechanic? I don't mean the wierd dice, I mean the base mechanic.

-clash