Dice Bucket Engines - *why*?

[Ancalagon]

Maybe I didn't express myself clearly enough:
If you're increasing a skill to increase your dice pool, say from 5 to 6, then you can suddenly roll as many as 6 successes. I.e. the range of possible outcomes has increased from 6 to 7. How would you do that with a single roll of percentile dice without requiring a table or matrix?

Nah, I wasn't clear enough. To answer your question: With ease.

Bob the peasant is hitting a goblin with his club. He has a weapon skill of 31%. (ie he needs to roll 31% or less to succeed... Bob isn't very good). In the warhammer system, a degree of success was defined as hitting your target by 10%. So Bob could have up to 3 degrees of success here.

(I'll note that degrees of success was *not* used in combat in warhammer, but for other things, but a combat example is easier).

Some time later, Bob is no longer a mere peasant but an experienced warrior and his weapon skill is now 51%. Not only that but he has been blessed by Fate, and has a magic sword! It improves his skill by a further 10%. Therefore, Bob can now hit the goblin with up to 6 degrees of success.

See? No need for a table or matrix.

 

[The Crimson Binome]

Some time later, Bob is no longer a mere peasant but an experienced warrior and his weapon skill is now 51%. Not only that but he has been blessed by Fate, and has a magic sword! It improves his skill by a further 10%. Therefore, Bob can now hit the goblin with up to 6 degrees of success.

What's interesting to me is that this gives the same distribution of success margins as rolling the attack as binary and then rolling another die to determine the margin of success. If you have a 60% chance to hit, and every 10% success margin increases the damage by 1, then you could get the same distribution by just rolling the percentile as binary and rolling d6 to determine the damage bonus.

 

[Ancalagon]

What's interesting to me is that this gives the same distribution of success margins as rolling the attack as binary and then rolling another die to determine the margin of success. If you have a 60% chance to hit, and every 10% success margin increases the damage by 1, then you could get the same distribution by just rolling the percentile as binary and rolling d6 to determine the damage bonus.

Which in actuality is what warhammer 2nd ed does (well, damage is rolled with a d10, not d6, but mere details). The degrees of success in this particular system is only used for certain types of skill checks, not combat. But clearly they could be. I merely wanted to illustrate that degrees of success are not a dice-pool only feature.

 

[GMMichael]

Bob the peasant is hitting a goblin with his club. He has a weapon skill of 31%. (ie he needs to roll 31% or less to succeed... Bob isn't very good). In the warhammer system, a degree of success was defined as hitting your target by 10%. So Bob could have up to 3 degrees of success here.

Some time later, Bob is no longer a mere peasant but an experienced warrior and his weapon skill is now 51%. Not only that but he has been blessed by Fate, and has a magic sword! It improves his skill by a further 10%. Therefore, Bob can now hit the goblin with up to 6 degrees of success.

See? No need for a table or matrix.

Except this degrees-of-success system does not match the dice pool's degrees of success.

I'm guessing that Jhaelen counted 7 "outcomes" by referring to the number of Successes showing on any given roll of the pool. What's not so obvious is that each outcome does not have an equal chance of occurring, and 7 Successes is the least likely outcome. However, with the percentile dice, your odds of getting the highest number of degrees of success are the same as getting a success with zero degrees.

This difference is so stark that I would not call both features "degrees of success." The success/fail pool is closer to degrees of success; I'd say the high-rolling-percentile is more like a random result.

PS I'd use a Success/Fail pool up to, oh, 6 dice. A Total Result Pool, probably the same. After that, I'll start to wonder whether I could be role-playing instead of counting/adding. However, there is a nice tactile reward to adding a die to your pool when it's small. I think the law of diminishing returns applies to that reward, though.

 

[pemerton]

[MENTION=6775031]Saelorn[/MENTION] True, but if that is a concern you can replace the d20 with 3d6 (or something like that). Probabilities around the average can be achieved without a dice pool no?

Over the past year I've run 4 systems, each of which uses a different dice engine:

4e D&D - uses d20 + adds

Burning Wheel - uses a dice pool as you have defined it in this thread - roll d6, count successes (generally 4+), try to meet a target number (the typical pool consists of skill/attribute plus situational modifiers)

Classic Traveller - mostly uses 2d6 + adds

Marvel Heroic/Cortex Fantasy - uses a fist-full of dice put together like a classic dice pool (one from each pertinent attribute/trait), but only a limited number (by default, 2) are added together to get the result, and a third die generates the degree of effect (eg damage or augment) based solely on its size, not its result​

Some of the differences I have noticed:

In 4e, auto-success becomes possible, especially for higher level PCs.

In BW, failure is always a possibility because no matter how many dice you roll, they might all come up 3 or less. Surprising failures are more common than in 4e (making it a grittier system). Another reason failure is more common is because buffs don't scale as quickly as penalties (adding a die adds, on average, half-a-sucess; whereas adding 1 to the target number requires an extra success) - the system leverages this, and the failures it leads to when challenges are hard, to produce a dramatic cycle that is quite different from D&D (it is hard, in a "linear" system like d20, to both (i) make success possible, yet (ii) make it really quite unlikely).

In Traveller, the 2d6 has a "smoothing"/averaging effect compared to (say) rolling a d10. Also, because skill bonuses tend to be fairly low (due to the vagaries of the PC generation system, even a +3 is rather unusual), the "flattening" is exaggerated, making higher difficulties harder to reach. Traveller, to me at least, feels closer to BW than to D&D, and I think the dice mechanics are part of that.

MHRP, at least in my experiences, produces the most "stylised" results, which normally are contained in a relatively narrow range of effects. Adding an extra die increase the range of results to choose from, but doesn't directly boost the result; that is achieved via a different dimension of the game, namely, spending "plot points".

 

[Ancalagon]

Except this degrees-of-success system does not match the dice pool's degrees of success.

I'm guessing that Jhaelen counted 7 "outcomes" by referring to the number of Successes showing on any given roll of the pool. What's not so obvious is that each outcome does not have an equal chance of occurring, and 7 Successes is the least likely outcome. However, with the percentile dice, your odds of getting the highest number of degrees of success are the same as getting a success with zero degrees.

This is quite obvious.

This difference is so stark that I would not call both features "degrees of success." The success/fail pool is closer to degrees of success; I'd say the high-rolling-percentile is more like a random result.

This is because warhammer, like d20, uses a "flat probability" dice (a % instead of a d20, but still each number is equally likely). If you want a system where average results are more likely than high/low ones, as I mentioned above, you can use 3d6 like Gurps or Hero does (or 2d6, 2d10 etc etc).

So it seems that you don't need a dice pool to get degrees of success, you don't need a dice pool to get a non-flat distribution of results, and you don't need a dice pool to get both. That being said, I do appreciate the effort that people are putting in answering my question!

(BTW, another example of a degrees of success system (although one with a flat dice, a d20). In star wars saga, when firing with a weapons battery (a cluster of vehicular weapons, not individual weapons), for each 3 above the target AC, one dice was added to the amount of damage the guns did.)

 

[pemerton]

So it seems that you don't need a dice pool to get degrees of success, you don't need a dice pool to get a non-flat distribution of results, and you don't need a dice pool to get both.

But you do need a dice pool to have a reducing but non-zero chance of failure regardless of skill level relative to target number. (Open-ended systems can approximate to this, but overall are closer to d20 attacks/saves with auto-fail on 1.)

 

[Ancalagon]

But you do need a dice pool to have a reducing but non-zero chance of failure regardless of skill level relative to target number. (Open-ended systems can approximate to this, but overall are closer to d20 attacks/saves with auto-fail on 1.)

I must admit I'm not sure if I understand correctly... Could you give me a an example or two?

Thanks!

Sent from my SM-G930W8 using EN World mobile app

 

[The Crimson Binome]

I must admit I'm not sure if I understand correctly... Could you give me a an example or two?

Imagine you're playing Exalted, and you want to craft a non-magical sword. This requires one success on a test, and as an exalted character who cares about crafting swords, you're probably rolling at least ten dice. Since each die has a six-in-ten chance of not generating a success, and you only need one success, your chance of failing to craft that sword is ~0.6 percent. If you were just some chump human, and you were only rolling five dice, then your chance of failing would be a whopping ~8 percent. As you get better and your die pool increases, your chance of success approaches 100% but never actually gets there.

Unless you alter the number of dice involved, there's no way easy way to reduce the chance of failure beyond a certain point, because you're limited by the size of the dice. In a d20 system, you can either design a zero percent minimum chance of failure or a five percent minimum chance of failure. If you're rolling 3d6, then you can either design a zero minimum chance of failure or a 1/216 minimum chance of failure. There's no nuance where a master can continue to get better and thereby reduce their failure chance, but with harshly diminishing returns.

 

[Ancalagon]

I agree with you that the 5% chance (or none) of the d20 is a bit harsh. However, the value of going form, say 1/216 to say 1/1000 is... dubious. It is an advantage, but a very minor one.

Sent from my SM-G930W8 using EN World mobile app