Dice Pool Mechanic vs. Single Die

[jester47]

Just got done looking over the skill mechanic for Burning Wheel.

One of the big points that I see in the marketing lit for BW is the idea that difficulty numbers are determined by the player and not the GM.

This is a bunch of baloney. Here's why:

A die pool with x number of dice that need to get over y target number to register against a z obstacle number breaks down to the mechanics of 1 die vs difficulty number mechanic.

3 dice that need to roll over a 3 have a graded % chance of getting a certain number of successes. When you compare successes against a number set by the GM, you are in essence reverting back to a die roll vs an arbitararily set challenge number.

The fact that the system contains the "obstacle" number just puts it back in the same place as d20.

The lesson is that in any RPG at some point someone is going to have to make a decision about how hard a task is to acomplish.

However I will admit that the player has MORE control over how successful he is. But that seems to be covered just as well with skill points. I prefer the statistical simplicity of the d20.

 

[AuraSeer]

This sounds like an interesting statistical comparison, but I'm not familiar with the Burning Wheel system. Could you post a short description of how it works?

 

[Thanee]

Well, you can do every system in exactly the same way, by using a d% and converting the required rolls into percentage chances. Still, not every system uses this most straightforward dice method.

Bye
Thanee

 

[Wil]

Note that this is specific to SilCore, and not the Burning Wheel - although they have similar mechanics.

In SilCore, you roll a number of dice equal to your skill level, keep the highest roll, and add modifiers. Unskilled characters roll 2 dice and keep the lowest; stat checks roll 2 dice and keep the highest; skill levels are a range of 1 to 5 (or 1 to 10, depending on some options that can be chosen for the game you're running). Rolling multiple 6s means you add +1 for each additional 6 (so a roll of 6,6,6 is read as 8); rolling all 1s is a fumble.

The advantage here over straight single die systems (roll d20 and add modifiers for example) is that unskilled characters are less predictable, while more experienced characters are virtually guaranteed a minimum level of performance. A character with skill 1 has the exact same odds of rolling a 1 or a 6. A character with a skill of 3 is more likely to obtain a result of 4 than any other result. To me, this type of performance closely models my own experience in the real world.

Can you emulate this with a single die system? Sure, but not usually without adding some complexity to how the dice are read. It's easier to let the results curve take care of that.

 

[arscott]

Will, that sounds more like a kept dice system, wherein multiple dice are rolled and the result is the sum of the numbers showing on one or more of the highest dice. The "3 dice that need to roll over a 3" sound more like a dice pool system (aka success system), wherein multiple dice are rolled and the result is the number of dice showing a certain result or better.

The first system is nice because it can create interesting probability curves, where the range of results depends on the number of dice you keep, but the chances of getting different numbers within that range depend on how many dice you discard.

The problem with the second system, is that it takes a whole bunch of extra dice to get a result that can just as easily be arrived at by the kept dice system or straight xdx rolls, but takes more dice and more math to do so.

Consider the following:
A conflict resolution system uses the following method: Roll four 4-sided die. A roll of 1-3 is a failure, and a roll of 4 is a success. Total the number of successes and compare to a target number.

Here's the likelyhood of each result using the dice pool system:
0 sucesses- 42.1875%
1 sucess- 31.640625%
2 sucesses- 21.09375%
3 sucesses- 4.6875%
4 sucesses- .390625%

Now compare it to this kept dice system:
Roll two 4-sided die. Keep the lower result and compare it to a target number.

Here's the likelyhood of each result using the kept dice system:
result of 1- 43.75%
result of 2- 31.25%
result of 3- 18.75%
result of 4- 6.25%

If you simply add one to your target number, you can easily swap out the kept dice system for the dice pool system. You use half the dice, and the calculation is much simpler. To me dice pools seem like a half-baked attempt to make the system seem more unique or interesting than it actually is, or possibly an attempt to make extra cash by encouraging people to buy official Vampire: The Masquerade dice sets. Either way, I'm not impressed.

Of course, I haven't actually seen Burning Wheel. Those mechanics might be different. This is just my rant at dice pool systems in general.

 

[Wil]

Well, it helps that SilCore was designed as a tactical wargame as well as a roleplaying game, and I'm pretty certain that the designers thought it out when they designed it.

For a character with skill 3 (roll 3 six-sided dice and keep the highest) the probability is nearly 35% that the result will be exactly 6, followed up by a nearly 30% chance the result is exactly 5. 2/3 of the possible results will more than likely be a 5 or 6. There's less than a 1% chance that character will fumble (and, likewise, less than a 1% of chance of achieving a roll of 8). The result set may seem very small to some people, but my experience in the real world is that there isn't a lot of gradiation between levels of success or failure anyway.

 

[painandgreed]

If you simply add one to your target number, you can easily swap out the kept dice system for the dice pool system. You use half the dice, and the calculation is much simpler. To me dice pools seem like a half-baked attempt to make the system seem more unique or interesting than it actually is, or possibly an attempt to make extra cash by encouraging people to buy official Vampire: The Masquerade dice sets. Either way, I'm not impressed.

I think this brings up the most important feature of game design, playability verus realism (or verisimilitude as this board likes to say). Make a game too relaisitic trying to similate the way things work in the real world too well, and it gets complicated and nobody wants to play it. There are many games out there whose systems are considered good but overly complicated to play and thus they die. Make it too easy, and the challenge and suspention of disbelief disappears and it fails to capture the attetion of the player.

Using half the dice may seem better but it allows for less differentiation between levels of skill. It becomes hard to accont for a wide range of skill levels. You can modify the target number but you increase the complexity of the system and lose some of the ease you get for using less dice. Go with a straight dice pool and you can account for a large range of skills, but the players can end up rolling a multitude of dice. Use one die and it bcoems easy to do, but may not capture the sense of realism that a player may want.

There's no one answer. Some people like simple and some like complicated. Some people may like one or the other due to the mood or game they're playing. There is no choice for everybody for everygame.

 

[jester47]

Note that this is specific to SilCore, and not the Burning Wheel - although they have similar mechanics.

All my information came from the burning wheel web site.

The character build determines the number the dice have to roll to be counted as a success, and the number of dice rolled, yet the GM determines how many successes are needed to acomplish a task.

Might as well have skill ranks and a DC and roll one die.

 

[Wil]

I think this brings up the most important feature of game design, playability verus realism (or verisimilitude as this board likes to say). Make a game too relaisitic trying to similate the way things work in the real world too well, and it gets complicated and nobody wants to play it. There are many games out there whose systems are considered good but overly complicated to play and thus they die. Make it too easy, and the challenge and suspention of disbelief disappears and it fails to capture the attetion of the player.

Using half the dice may seem better but it allows for less differentiation between levels of skill. It becomes hard to accont for a wide range of skill levels. You can modify the target number but you increase the complexity of the system and lose some of the ease you get for using less dice. Go with a straight dice pool and you can account for a large range of skills, but the players can end up rolling a multitude of dice. Use one die and it bcoems easy to do, but may not capture the sense of realism that a player may want.

There's no one answer. Some people like simple and some like complicated. Some people may like one or the other due to the mood or game they're playing. There is no choice for everybody for everygame.

The thing is, I don't believe that there are a wide range of skill levels that people have. I feel it can be best represented by a half dozen or so broad levels, and that you really can't differentiate between fine levels of performance. Either someone does a job poorly, they do it adequately, they do it well, or they do it perfectly. I'm that way about the range of human ability - I just don't think there are fine variations in the range of, say, manual dexterity. Either you suck at it, you're average, you're good or you're really good. This is why SilCore, with the default range of attributes from -3 to +3 (although Silhouette originally went from -5 to +5) and 5 skill levels (6 counting unskilled) suits me just fine - because the base results meet my expectations of reality. I can easily manupulate those results using rules options to better simulate more dramatic (or cinematic) styles of play. A lot of it does boil down to player preference, but a good resolution system is going to a) provide a reasonable mathematical base for determing success or failue and b) be scaleable, or at least modular enough, to be able to accomodate varying levels of "realism".

If you look at a straight d20 + modifier system, it can be said to fail in at least one of those criteria. There is the same chance of rolling a natural 1 or 20 regardless of skill levels. With bonuses it may not be possible for a character to get below a certain roll, meaning that if you want "fumble" mechanics you need to say, "If you roll a 1, no matter what your bonuses are, it's still a failure". Some people might say that a constant 5% failure rate is just too high in many circumstances (BTW, SilCore works around this by saying that if you roll all 1s it is a fumble, but you can still succeed at the task if the roll after bonuses dictate. So if you get a final result of 4 after rolling all 1s with a gun and it's enough to hit, you hit but your gun jams). It's one of the reasons that I dislike single die systems...even roll 3d6 and total the results (ala GURPS) is preferable because there is some amount of a curve. Note that GURPS does have more complex criteria to deal with minimum failure rates - the lowest roll to be considered a fumble is dictated by the skill level, so someone with a high skill needs a lower roll to fumble than someone with lower skill. Roll and keep systems are able to streamline this by moving the operation of "compare roll to value determined by skill level" into the actual die roll.

Now that I think about, that would seem to me to be the ultimate goal of a die roll mechanic - to incorporate as much "processing" into the roll itself, and reduce special cases of how to read the die and modifications to the final roll as much as possible.

 

[Wil]

All my information came from the burning wheel web site.

The character build determines the number the dice have to roll to be counted as a success, and the number of dice rolled, yet the GM determines how many successes are needed to acomplish a task.

Might as well have skill ranks and a DC and roll one die.

As I pointed out, one die is exceptionally poor at creating a consistent base of performance.

Rroll 3d6 and keep the highest will result in a roll of 4-6 75% of the time, .5% chance of rolling all 1s, and varying probabilities of rolling any other number.

Rolling 1d20 will result in a roll of 11-20 50% of time, and a straight 5% chance of rolling any one other number. It seems like 1d20 is more predictable, but it's really not - your highly skilled character when deprived of bonuses is just as likely to be nerfed by a bad roll as a good one.

In SilCore, the skill level 3 veteran with no attribute bonuses will more consistently beat the skill level 1 hotshot with an attribute bonus more often than not - which brings up the other reason that a roll and keep system like this works so well. It allows for those random lucky breaks. Sure, the veteran will be consistently rolling 5s and 6s and putting the beat down on the rookie, but at some point that rookie is going to roll a 6, add his bonuses in, and get a lucky shot in. In single die systems, this either happens more often than is reasonable or can never happen at all.

It also makes it difficult to implement automatic fail/success mechanic. If you say that rolling a 1 is always a fumble, then you have just created a situation where all tasks are fumbled 5% of the time.

There's a very good reason to have the mechanics set up that way, it's just not a newsletter that you subscribe to.

I'd also like to note that while SilCore does not count successes, it does have both a difficulty level and a complexity rating for each task. The roll must beat the difficulty level, and the character must meet the complexity rating with his skill in order to not be penalized. It's by far my favorite resolution system because it offers multiple axises to evaluate tasks. The GM still sets the threshold and complexity of the task, but there are mechanics in place to allow players to accomplish more and varied effects if they want to sacrifice skill levels or the complexity they have in the skill exceeds what is needed to perform the task.