Is it even possible to have reasonable dicepools for D&D?

[Lago PARANOIA]

Soooo... here are my criteria.

1) You use the Shadowrun-style dicepools. You don't have to keep the TNs static, but the DCs have to be.

2) The RNG models a range of results from 'can keep balance across an oil-covered linoleum floor' to 'can balance on water'. As in stuff you'd see in a low-grade fairytale to things that you'd see in, say, Naruto.

3) The RNG is constructed in a way to effectively lock out people unless they roll enough dice. No peasants getting a result of 'climb rain'.

4) The dicepool is kept at a reasonable size. Anything more than 15 dice rolled is way too much and probably even 12 dice.

Is such a thing possible? The only trick I can think of is to shift TNs around and/or give extra hits for rolling the top end of the die (meaning that you use d12s or d10s) or just being at a certain level. By inspection this seems to be an impossible task, but I'm not a math guy.

[sabs]

the Shadowrun DP system has some serious bellcurve and granularity issues. I'm not sure it's really an improvement.

[Username17]

Dicepools can model exponentials to a limited degree. They scale faster and more evenly when the target number on each die is lower. When the TN is higher, people need to roll more dice to gt anything done and people infrequently roll a lot higher than expected. With a Target Number of 3, the maximum result is only 50% more than the average result, which makes it a lot easier to cut off peasants from climbing rain.

-Username17

[RobbyPants]

Yeah, that sort of reminds me of the Betrayal at House on the Hill dice. They're all d3 -1 (0 to 2), so when you roll X dice, you always have a minimum of 0, an average of X, and a maximum of 2X. If you keep enough reigns on the number of dice people can roll, it's not that hard to lock out certain abilities.

[Lago PARANOIA]

the Shadowrun DP system has some serious bellcurve and granularity issues. I'm not sure it's really an improvement.

I'd still rather use a 3d6 or a d20 because it's easy to figure out what's going on, I just want to know if it can be done.

With a Target Number of 3, the maximum result is only 50% more than the average result, which makes it a lot easier to cut off peasants from climbing rain.

Right, but won't that cause the dice pool to inflate too quickly?

[CCarter]

I don't know offhand of any level-based success counting dicepool systems.

If your dicepool is [stat + level in a particular skill] like Storyteller say, that's only up to 5 dice from level advancement to get to a limit of 10 dice.

The range of your dice pool increases if your stat does something other than grant extra dice, I suppose: automatic successes perhaps?

There are also some workarounds for huge dice pools: you can work out the probabilities, convert to a table and roll d100 (or something, depending on your rounding tolerance) on it for characters who are breaking the sensible # dice limit.

Or if the TN gives you a 50/50 success chance (and no roll 10s again or whatever) you can compress your dice pools by using custom dice, like Ubiquity Dice in Hollow Earth Expedition. The HEX people sell a d8 labelled 0,1,1,1,2,2,3 which is equivalent to the probability of rolling 3 ordinary dice and counting the hits, so you can compress the number of dice you roll to 1/3. A player effectively reads these dice by adding them all together rather than scanning for successes.

[Psychic Robot]

Nobody wants dice pools for D&D. If you want to make a fantasy heartbreaker with dice pools, go for it. But put "D&D" in the title and you'll be skewered for it.

[Sashi]

With a Target Number of 3, the maximum result is only 50% more than the average result, which makes it a lot easier to cut off peasants from climbing rain.

Right, but won't that cause the dice pool to inflate too quickly?

The required size of a dicepool to get any number of successes increases the higher the TN is. So if your system has a TN of 6 people will be throwing around dicepools of 6-18 dice just to reliably get 1-3 successes, with a chance at success rates 6 times higher than their average. While if the TN is 2 then the pools will be from 2-5 dice, with a chance at 1.2 times their average successes.

[Lago PARANOIA]

Yes, but another goal of the system was to cut peasants off from getting weird results like 'successfully forage for food in Pandemonium' or 'casually intimidate Titans'. Reducing the TN will reduce the number of dice you need to get a result, but won't it admit more mooks into getting 'awesome' results?

Or if the TN gives you a 50/50 success chance (and no roll 10s again or whatever) you can compress your dice pools by using custom dice, like Ubiquity Dice in Hollow Earth Expedition. The HEX people sell a d8 labelled 0,1,1,1,2,2,3 which is equivalent to the probability of rolling 3 ordinary dice and counting the hits, so you can compress the number of dice you roll to 1/3. A player effectively reads these dice by adding them all together rather than scanning for successes.

Wow, that sounds amazing. Do they make any other kinds of dice?

[CCarter]

Just those I think, in a couple of varieties.
Review of the things:

http://www.rpg.net/reviews/archive/12/12634.phtml

Maker's website
http://www.exilegames.com/access/dice.html

A little expensive is maybe the drawback. I hadn't realized but they only seem to sell these in tubes of 9 dice - 3 'd1s' (regular d8 with half the sides 1 and half 0 i.e. not so great), 3 'd2s' (replace roll of two dice) and 3 'd3s' (which replace the three dice as noted above). A tube is 5.99 US.

[Manxome]

Yes, but another goal of the system was to cut peasants off from getting weird results like 'successfully forage for food in Pandemonium' or 'casually intimidate Titans'. Reducing the TN will reduce the number of dice you need to get a result, but won't it admit more mooks into getting 'awesome' results?

No...as Frank and Sashi both already explained, lower TNs moves the average and the maximum closer together, so it does the opposite of that.

For example, suppose you use a TN of 3, and assign dice pools like this:

	Dice Pool	Avg Hits	Max Hits
Peasant	3	2	3
Hero	6	4	6
Specialist	9	6	9

The non-specialist hero has a >50% chance of doing 4-hit tasks, which the peasant cannot possibly do on any roll ever, and has about a 9% chance of doing the things that the specialist usually succeeds at (6 hits), with no chance at all of doing things that are hard for a specialist. And all the numbers are single-digit.

[erik]

Wow, that sounds amazing.

Oh sure you say that now. Nobody paid it mind over a year ago. *pout*

A little expensive is maybe the drawback. I hadn't realized but they only seem to sell these in tubes of 9 dice - 3 'd1s' (regular d8 with half the sides 1 and half 0 i.e. not so great), 3 'd2s' (replace roll of two dice) and 3 'd3s' (which replace the three dice as noted above). A tube is 5.99 US.

At cons they seem to sell them for $5 per tube. I got 2 tubes and that seemed enough for my whole table. One tube for players one tube for the MC (me at the time). $10 (or $12) for dice for a game wasn't too harsh I thought.

[CCarter]

*hangs head*
Sorry Erik. Come to think of it...I think I did first hear about HEX here somewhere, though in an offhand mention (defensive skill use maybe?). I don't recall seeing the full review thread, though.

[erik]

Hah, nothing to be sorry for. I used to mention it more often from late 2009 to intermittently in 2010, but nobody really paid it any heed, perhaps because I'm a poor shill. I shoulda shown pictures of the dice

It was a fun system to use despite its lack of attention to game balancing (mostly when it comes to the Talents, though other problems do remain as well). My group had a lot of fun with it, but for the last couple years we've had a really difficult time setting up an RPG schedule on account of a few of us (myself included) being new parents, along with other distractions and flakiness.

It's too bad that Exile Game Studio's release schedule was/is so slow. If they had a few more (preferably math-inclined) writers and an aggressive schedule a few years ago then I think they could have become something big. The moment was there for another RPG crew to fill a void. 2 pdf Adventures released on DriveThruRPG in 2010-2011. One supplement in 2009 and another in 2008. It's too bad.

[Shazbot79]

I wonder if anyone has ever thought to write character options granting automatic successes to certain rolls.

I briefly considered a dicepool-based game wherein players rolled a handful of d6's, counting successes against a target "threshold" of 1 to 10, with 4's and higher on the dice counting as a success. The idea was to ponder if a stripped down shadowrun is feasible.

Enhanced attributes like "Super Strength" or "Lightning Reflexes" would grant automatic successes instead of adding to the dice pool. This way, the dicepools wouldn't stretch past 10.

My experience with dicepools games is limited, so I'm not 100% convinced that this would work in actual play, but I remember drawing up the probabilities and they seemed to work (with the assumption that a difficulty 10 challenge is a feat of epic cosmic power that even specialists might have trouble with). I wonder if anyone has tried this before?

[violence in the media]

oWoD Potence granted automatic successes on Strength related challenges, regardless of what the TN was. This included weapon damage as well.

[erik]

I wonder if anyone has ever thought to write character options granting automatic successes to certain rolls.

In HEX/ubiquity games you are free to do so whenever. For any even number of dice you can take the average, on odd numbers take (average of n-1 of dice)+1d2. Really speeds things up when desired.

[Username17]

I wonder if anyone has ever thought to write character options granting automatic successes to certain rolls.

In Shadowrun and aWoD, you average one hit per three dice. But you can buy hits to automatically get a level of success at the rate of one hit per four dice.

-Username17

[Judging__Eagle]

If dicepools were meant to simulate difference in tier somehow, and were capped at a smallish number (say 1d6 if you have an attribute or ability, and 5d6 for max; upgradable to 1d3 and 1d6 which generate "hits" based on dice roll at roughly double the cost of the dice bought) for comparisons within a tier.... then you might have something like D&D, where mooks never threaten heroes; but on-par, or greater-than-par, enemies are still a danger to PCs.

The d20 would be used a lot less, only when there is a massive discrepancy between an actor and their target; and they would generated 1d20 successes on the roll.

Most often you'd see a d3 or d6 in successes when there is a discrepancy in power.

Of course, I'm stuck on the idea that games tend to have "tiers", and that D&D has very dramatic and noticable breakpoints between tiers of power that I can use as guidelines, breaking them up into "Mortal" (1-5), "Heroic" (6-10), "Legendary" (11-15) and "Epic" (16-20).

[Lago PARANOIA]

So, here's the situation.

I want to have a resolution mechanic that uses d6s if possible, since they're the most common dice. I'm open towards having the TN shift.

I want to have approximately 8 distinct ranges of skill. Each division is a level of specialty.

Peasant
1st-Level Generalist Hero
1st-Level Specialist / 3rd-Level Generalist
3rd-Level Specialist / 6th-Level Generalist
6S / 9G
9S / 12G
12S / 15G
15th-Level Specialist.

I also want to have three broad categories of skill advances: Sharply-Tiered Skills, Average-Tiered Skills, and Flatly-Tiered Skills. For Sharply-Tiered Skills, Something that's Average difficulty (50% chance) for someone is Very Difficult/Near Impossible (5-15% chance) for someone who is one level below to do and Very Easy (85-95% chance) for someone one above. Examples of Sharply-Tiered Skills: Athletics, Acrobatics, Spellcraft, Computer Use.

For average-tiered skills, something that's average for someone should be Very Difficult/Near Impossible to someone TWO levels of specialty below and Very Easy for someone two above. Examples: Stealth, Perception, Knowledge, Survival.

For Flatly-tiered skills, something that's average for someone should be Very Difficult/Near Impossible to someone that's THREE levels of specialty below and Very Easy for THREE above. Examples: Craft, Sense Motive, Diplomacy, Profession.

Examples by average difficulty:

Specialty Level	Sharp (Athletics)	Average (Stealth)	Flat (Diplomacy)
1	Climb Tree	Search sleeping guard without waking.	Convince merchant to give you discount.
2	Climb Stone Wall	Hide from angry, searching street mob.	Guards not to arrest you immediately despite being top suspect for heinous crime.
3	Climb Smooth, Oiled Stone	Sneak into lion cage and steal cub while lions are awake.	Hanging judge to show large amounts of mercy.
4	Climb Marble Ceiling	Walk past several extremely attentive guards in plain sight.	Centuries-old previously unrepentant vampire to earnestly look for cure to condition.
5	Climb Waterfall without using Rocks	Stand directly in front of security cameras and act for several minutes without being seen.	Leopold II to halt genocide and sincerely make amends for what he did in Congos.
6	Climb a geyser of steaming blood	Trick out an effect that uses your truename.	Ravagers to disband, forswear god of slaughter, rejoin civilization for punishment.
7	Climb a light drizzle of rain.	Avoid disturbing a magnetic field while walking through a km of it in a Faraday cage.	Hades to release an extended family's souls.
8	Climb horizontally contrariwise to a tornado	Hide soul from direct gaze of several gods of death without taking it from body.	Ares to sue for peace and enforce it.

So what should be the size of the dicepools?

[Lago PARANOIA]

Bonus Project:

If you wanted to have a Black Forest game where things more or less transitioned from Peasant tier to 1st-Level Specialist but had, say, 5 discrete units of advancement, how would you implement it?

For example, here would be the specialty levels:

1 Small Child with no training, only vaguest idea how to do things. You're no better at this than a toddler.
2 Town Drunkard, Bright but unskilled prepubescent kid.
3 Bog-Standard Peasant living on mud farms or farting around in a city.
4 Squire
5 1st-level 'regular' hero, like a Paladin, Wizard, etc.

The TNs and dicepool of the 'regular' hero and peasant need to be the same as those for the standard D&D project, but anything below that is okay. For example if the TN for the rest of the system is 3 and peasants have a dicepool of 3, I will totally accept small children having a dicepool of three die but a TN of 5.

[Manxome]

For Sharply-Tiered Skills, Something that's Average difficulty (50% chance) for someone is Very Difficult/Near Impossible (5-15% chance) for someone who is one level below to do and Very Easy (85-95% chance) for someone one above.

That is not how dice pools work. If you want a system where adding one tier always shifts a 50% chance to a 90% chance and a 10% chance to a 50% chance, regardless of what tier you were at before, your tier needs to be adding a static number to your result, NOT adding dice or adjusting TNs.

Changing the TNs or the number of dice changes the shape of the curve, not just the expected result.

If you want, you could have a system where everyone rolls X dice with TN Y for all things all the time, and you add a static number of hits to the final result based on your tier. But a system where increasing in tier adds dice to your pool is fundamentally and completely incompatible with your stated goal.

For average-tiered skills, something that's average for someone should be Very Difficult/Near Impossible to someone TWO levels of specialty below and Very Easy for someone two above.
...
For Flatly-tiered skills, something that's average for someone should be Very Difficult/Near Impossible to someone that's THREE levels of specialty below and Very Easy for THREE above.

And that's just saying that all "sharp" bonuses should be twice as big as "average" and three times as big as "flat" bonuses.

So here's an example of approximatley what you asked for: everyone rolls 1d20 + bonuses. For a "flat" skill, your bonus is +3 per tier. For an "average" skill, it's +4 per tier. For a "sharp" skill, it's +8 per tier.

Thus, for a "sharp" skill, something that's currently a 50% chance of success will become 10% if you drop one tier or 90% if you go up one tier. For an "average" skill, you need to change two tiers for the same effect; for a "flat" skill, you need to change 2.67 tiers (I rounded).

If you want a bell curve and/or to use six-sided dice, you can roll 3d6 and use something like +1, +2, and +4 for the tier bonuses.

You could also have a dice pool system where everyone rolls the same number of d6s against a TN of 4 (a TN of 4 is mandatory if you want symmetry, which you claim you do), and then adds a static number of hits based on their tier, but to get the spread you asked for using only integers would require a very large number of dice. In fact, in order to get the exact relationship between tiers you asked for, the "sharp" bonus per tier needs to be a multiple of 6 (so that "half" and "one third" of that are both whole numbers), which means that 6 needs to be less than half your result spread, so you'd be looking at a minimum of 13 dice, and a +6 on 13 dice TN 4 changes a 50% to a 99.99%, not to a 90%.


If none of the above is satisfactory, you probably need to change your requirements.

[Lago PARANOIA]

That is not how dice pools work. If you want a system where adding one tier always shifts a 50% chance to a 90% chance and a 10% chance to a 50% chance, regardless of what tier you were at before, your tier needs to be adding a static number to your result, NOT adding dice or adjusting TNs.

This is why I gave a fuzzy range. I wanted to convey the whole 'not totally impossible but not better than an outside last-ditch chance' (I think I said like 5%-15% chance, but that's too low I feel), which is why I left the middle possibilities for the flatly-tiered and average-tiered curves blank. I'm okay with psuedo-symmetry, as long as players have a general idea about how likely they are to succeed. IOW it's okay for 'outside chance but not impossible' to be 25% for one dicepool array and 10% for another, but not 1-5%.

[Manxome]

Let me put that another way: if what you want is a system where a relative difference of 1 tier means basically the same thing no matter where you are in the hierarchy, why would you want to be using dice pools in the first place?


But OK, let's run some numbers. Let's suppose we only use one of your criteria (say, going up one tier changes the easiest task with <15% success to >=50% success), then fix the TN at 5, give 2 dice to tier 1, and see how dice pools grow?

Tier 1: 2 dice. 56% at least 1 hit, 11% 2 hits
Tier 2: Need to get 2 hits >=50% of the time, so we have 5 dice. 54% at least 2 hits, 21% 3 hits, 5% 4 hits
Tier 3: 11 dice. 53% at least 4 hits, 29% 5 hits, 12% 6 hits
Tier 4: 17 dice. 52% at least 6 hits, 17% 8 hits, 8% 9 hits
Tier 5: 26 dice. 52% at least 9 hits, 22% 11 hits, 12% 12 hits
Tier 6: My current probability table only goes up to 30 dice.

Now check the opposite condition: what's the odds of hitting the 50% hit threshold from one tier further up:

Tier 2 attempting Tier 1 normal: 87%.
Tier 3 attempting Tier 2 normal: 92%
Tier 4 attempting Tier 3 normal: 87%
Tier 5 attempting Tier 4 normal: 91%

That's roughly in line with your targets, though of course you're going to be rolling a completely unwieldy number of dice very quickly.


We said earlier than lower TNs are better for shutting low-tier people out. Let's try a TN of 3:

Tier 1: 2 dice. 89% at least 1 hit, 44% 2 hits, 0% 3 hits
Tier 2: Need to get 3 hits >= 50% of the time, so we have 4 dice. 59% at least 3 hits, 20% 4 hits, 0% 5 hits
Tier 3: 7 dice. 57% 5 hits, 26% 6 hits, 5% 7 hits
Tier 4: 10 dice. 56% 7 hits, 30% 8 hits, 10% 9 hits
Tier 5: 13 dice. 55% 9 hits, 32% 10 hits, 14% 11 hits
Tier 6: 16 dice. 55% 11 hits, 17% 13 hits, 6% 14 hits
Tier 7: 21 dice. 60% 14 hits, 25% 16 hits, 12% 17 hits
Tier 8: 25 dice. 54% 17 hits, 11% 20 hits

Tier 2 attempting Tier 1 normal: 99% 1 hit, 89% 2 hits
Tier 3 attempting Tier 2 normal: 95% 3 hits
Tier 4 attempting Tier 3 normal: 92% 5 hits
Tier 5 attempting Tier 4 normal: 90% 7 hits
Tier 6 attempting Tier 5 normal: 87% 9 hits
Tier 7 attempting Tier 6 normal: 94% 11 hits
Tier 8 attempting Tier 7 normal: 91% 14 hits

That's a little more reasonable. It's kind of wonky on the low end (tier 1 and tier 2 have no chance at all at an "average" task of the next tier), and the steps aren't uniform (though closer than I expected), but its roughly in line with what you asked for and a tier 8 sharp skill is rolling "only" 25 dice, which is actually less than I expected.

Tier 1 still only has 3 possible outcomes for any roll, of course. You could move everything up a step, starting tier 1 at 4 dice, which will push tier 8 up to 29-30 dice.

The early steps are only going up 2 dice per tier, which means you can't fit a "flat" skill onto this same scale. You could try putting "average" on a TN 4 table and "flat" on a TN 5 table, and the tables wouldn't look anything alike but they might all have vaguely sane precision and numbers of dice (provided you're willing to roll 20 or 30 dice at the high end).

[Lago PARANOIA]

30 dice might be a little too much. You mentioned flat tier bonuses earlier, what kind of automatic hits would you have to assign to a d6 at different points in the game in order to keep the dice pool size to a maximum of, say, 16 dice? Also if people are allowed to 'buy' hits at the rate of 1 per 4 die would would be the expected average in combination with this system?

I don't know how much work it is to construct this RNG. Regardless, thanks for your help.

Let me put that another way: if what you want is a system where a relative difference of 1 tier means basically the same thing no matter where you are in the hierarchy, why would you want to be using dice pools in the first place?

I don't want--say for an average curve--the absolute difference between tier 1 to 2 to be the same as 2 to 3. The only way I could implement this with a d20 was to make the bonuses and DCs exponential but that got out of hand pretty quickly.