Quote from: Eirikrautha on January 11, 2024, 06:48:26 AM

Quote from: Steven Mitchell on January 11, 2024, 12:01:28 AM

Quote from: Eirikrautha on January 10, 2024, 10:11:46 PM
First, I have considered using two different methods to manipulate the die rolls (and, yes, I've got spreadsheet after spreadsheet from anydice.com to prove it!).  I've considered having anything that gives you a bonus adding a die to your pool.  Anything that makes your task harder will increase the target number (so I'd probably start the target number at 5+ if I went this route).  Mathematically, increasing the target number has a larger effect than increasing the number of dice.  This means the GM can be a little more liberal with bonuses, because anything that adds a penalty will have an outsized effect (so a couple of extra dice won't really offset being prone and blinded, for example).  I know that folks have expressed concerns upthread about not being able to discern the effects of varying multiple features of the die rolls.  Do you think I can get away with this, or would it put you off or cause consternation?

It's not so much as not being able to discern the effects for me (though that is present to some degree) as once you introduce multiple ways to modify the pool, it vastly increases the confusion for casual players.  If the system is such that you don't expect to support casual players, then I wouldn't worry about that so much. With any player, I also find a slight handling time penalty for moving the target number.  Since handling time and casual players are both important to me, I'd never design a dice pool system with anything other than fixed target number and simply changing the dice as the means of scaling.

There is, however, one exception that I have flirted with for some mild success in handling large pools:  After a set number of dice, give average successes instead of rolling them.  To me, the max number to roll should be somewhere between 6 or 10 (depending on how much you want to focus on ease of use with a lower number or allowing the trend towards average with the higher number).  Not surprising, I settled on 8 for a max last time I tried it.  Since I conveniently was using d6s with 4+ as success, it was easy to say that every 2 dice you had over 7 instead gave you one automatic success.  Map that out in any dice, and it's not all that different than rolling it.  Of course, this kind of thing doesn't have to be exactly average either.  You can skew it in a way that fits the system.  D10s with a success on 7+, you might arbitrarily trade 3 dice over the threshold for an automatic success, cheating the character a bit.

That's a really good idea.  It also gives the players a sense of reward, in that they have swung the odds in their favor so much the dice can't completely screw them anymore.  I like it.  Thanks for the suggestion!

I've looked into on AnyDice after first encountering it as an option in WEG's d6 Space, but apparently once you hit about 5 rolled dice the variance above that between "Xd" and "5d+averaging the rest" is truly negligible.

So long as it kicks in at or above 5d, the number of times it'd vary outside that window is for practical matters nonexistent.

It amuses me because if the probability for a success on each die is 50% it's ALMOST the Fudge dice (the four dice with equal chances of +1, 0, -1 basically create the same sort of standard deviation around a set number).

There is nothing wrong with dice pools and I can think of at least two vaguely OSR examples:

- In BECMI, one option for ability checks is rolling 2d6/3d6/4d6/5d6 under ability (which I find a good idea because the numbers work for me).
See here:
https://methodsetmadness.blogspot.com/2016/12/roll-xd6-under-ability-yet-another-d.html

- In the new LotFP draft, it seems that saves are now dice pools (which I find a bad idea for different reasons).

There might be something wrong with some specific implementation, but basically you can get what you're looking for with dice pools: predictability, granularity, dice pools, rare results... just adjust how you use them.

The reason that they don't fit well with OSR (there are a few exceptions as mentioned) is because OSR is meant to be compatible with D&D products that are usually d20.

But if you really want it, it could be done IMO; you could start with using OD&D HD for combat, for example.

In general I really prefer dice pools over a single d20, both because they tend to be more granular in the outcome and because they can help avoid the d20's failing feature that 5% of the time Einstein fails a math check so badly that he impales himself with his slide-rule and dies.  Most d20 rulesets have no means to circumvent that 5% crit fail even if someone is a master at what they're doing, which is ridiculous to me.  I really like Fantasy Flight's Star Wars dice pool setup for a bevy of reasons:

* More granular than pass/fail.
* Resistance is treated as increased difficulty on the actor instead of the defender rolling their own attempt to dodge/block/whatever.
* High stats/skills upgrade dice rather than add dice, so you aren't rolling like twenty dice on a check even at higher levels.
* The presence of minion groups streamlines things so you aren't rolling ten times for a bunch of stormtroopers and bogging down the encounter.

The whole custom dice/counting icons bit is a flat negative for a lot of people, but I think the other aspects of that dice pool system are solid gold and could be built into a more traditional dice pool with standard dice and a target number.

Savage Worlds is also a pretty solid dice pool setup since it keeps the pool down to two dice.

Shadowrun is, well, I've been playing since it came out, so I have a soft spot for it.  It can definitely get too deep in the weeds sometimes in the name of simulationism, but I feel the pros outweigh the cons.  I know my opinion is somewhat in the minority there.

Quote from: Valatar on January 11, 2024, 02:47:28 PM
In general I really prefer dice pools over a single d20, both because they tend to be more granular in the outcome and because they can help avoid the d20's failing feature that 5% of the time Einstein fails a math check so badly that he impales himself with his slide-rule and dies.  Most d20 rulesets have no means to circumvent that 5% crit fail even if someone is a master at what they're doing, which is ridiculous to me.

That's not dice, and a die pool can have similar issues, even if in slightly different places.  That's a failure of GM adjudication, calling for rolls when success should be ruled automatic or impossible, and moving on without touching the mechanics at all.  Or in many cases, a failure of the game to teach GM adjudication in any kind of coherent fashion.

There are very few GM's, even horrible ones, that would call for a roll when a character walked slowly across level ground, with nothing else happening, and no time limit.  Or call for one to keep breathing.  Somehow, that basic fact fails to register as a general principle, though.

Quote from: Steven Mitchell on January 11, 2024, 02:55:10 PM
That's not dice, and a die pool can have similar issues, even if in slightly different places.  That's a failure of GM adjudication, calling for rolls when success should be ruled automatic or impossible, and moving on without touching the mechanics at all.  Or in many cases, a failure of the game to teach GM adjudication in any kind of coherent fashion.

There are very few GM's, even horrible ones, that would call for a roll when a character walked slowly across level ground, with nothing else happening, and no time limit.  Or call for one to keep breathing.  Somehow, that basic fact fails to register as a general principle, though.

If the examples are more in doubt, the issue still exists in any system that uses 'number of successes' rather than 'cumulative results'.

To wit, my introduction to the 2d20 system was Fallout game hosted by a guest GM.  I made a character specialized in treating injury, to the point of excluding everything else.  I failed my first and only Medicine check to treat a routine injury.  The unskilled player behind me made the roll easily.

It's a flaw.

One difficulty with adoption but not use is unfamiliarity.  I've been gaming for decades, run a lot of systems, played a few more, and I just realized I've never played a dice pool game.  I'm not against trying it, but when I'm looking at two systems to learn and one is a dice pool, I've always thought "maybe next time."  I'm sure it ain't rocket science, but figuring out a new dice mechanic when I'm already learning a whole new system is enough to tip the balance scale.

Quote from: Steven Mitchell on January 11, 2024, 02:55:10 PM
That's not dice, and a die pool can have similar issues, even if in slightly different places.  That's a failure of GM adjudication, calling for rolls when success should be ruled automatic or impossible, and moving on without touching the mechanics at all.  Or in many cases, a failure of the game to teach GM adjudication in any kind of coherent fashion.

There are very few GM's, even horrible ones, that would call for a roll when a character walked slowly across level ground, with nothing else happening, and no time limit.  Or call for one to keep breathing.  Somehow, that basic fact fails to register as a general principle, though.

I was exaggerating in my example, because of course there shouldn't be rolls to climb out of bed.  But in the d20 system say a level 20 rogue confronted with a DC 10 lock still has that 5% failure chance.  Or, if you go the scaling route of 4e or Pathfinder 2/Starfinder, the level 20 rogue will inexplicably only encounter DC 40 locks at that stage in their career as the entire world has been upgrading their security systems behind the scenes as the rogue has gained levels.

Quote from: Valatar on January 11, 2024, 04:12:44 PM

Quote from: Steven Mitchell on January 11, 2024, 02:55:10 PM
That's not dice, and a die pool can have similar issues, even if in slightly different places.  That's a failure of GM adjudication, calling for rolls when success should be ruled automatic or impossible, and moving on without touching the mechanics at all.  Or in many cases, a failure of the game to teach GM adjudication in any kind of coherent fashion.

There are very few GM's, even horrible ones, that would call for a roll when a character walked slowly across level ground, with nothing else happening, and no time limit.  Or call for one to keep breathing.  Somehow, that basic fact fails to register as a general principle, though.

I was exaggerating in my example, because of course there shouldn't be rolls to climb out of bed.  But in the d20 system say a level 20 rogue confronted with a DC 10 lock still has that 5% failure chance.  Or, if you go the scaling route of 4e or Pathfinder 2/Starfinder, the level 20 rogue will inexplicably only encounter DC 40 locks at that stage in their career as the entire world has been upgrading their security systems behind the scenes as the rogue has gained levels.

Still not seeing it.  Yeah, the whole 3.5/4E/PF crazy scaling is a problem trying to always have a mechanical solution to an adjudication problem.  So of course it doesn't work well.  There's no free lunch around this.  If in adjudicating you think there's no way that a level 20 rogue will fail against DC 10 locks under normal circumstances, then you rule that he doesn't need to roll. 

Now, if you are saying there should still be a chance of failure, but not 5%, then that is a granularity issue.  My answer however would be that if the chances are less than 5%, I'm probably going to rule no roll is needed.  So mechanics to handle 3% or 2% or 1% chances don't help all that much.  In any case, it's part of the baggage that comes with any system--finding the lowest level chance of failure that still requires a roll, then requiring a roll for it.

Quote from: Eirikrautha on January 10, 2024, 10:11:46 PM
First, I have considered using two different methods to manipulate the die rolls (and, yes, I've got spreadsheet after spreadsheet from anydice.com to prove it!).  I've considered having anything that gives you a bonus adding a die to your pool.  Anything that makes your task harder will increase the target number (so I'd probably start the target number at 5+ if I went this route).  Mathematically, increasing the target number has a larger effect than increasing the number of dice.  This means the GM can be a little more liberal with bonuses, because anything that adds a penalty will have an outsized effect (so a couple of extra dice won't really offset being prone and blinded, for example).  I know that folks have expressed concerns upthread about not being able to discern the effects of varying multiple features of the die rolls.  Do you think I can get away with this, or would it put you off or cause consternation?

Second, I am structuring the dice pools to average between 3 and 5 dice (d10s), usually with only a couple of bonus dice available.  I don't foresee any rolls being over 7 dice.  But I am looking at mechanics that might play into the number of dice, like losing dice each time you do something strenuous (and regaining dice at a rate each round you don't... sort of a fatigue or stamina mechanic).  Will varying the number of dice each round be too much, especially with varying target numbers?

I'm curious as to your reactions and reasons...

Let me speak as someone with experience tinkering with dice pools which are about as complex as they can get without breaking.

First, about players not being able to discern features; this is a half truth. The root problem is that you are trying to deliver game feel through the metaphorical rubber hose that is RNG. If your RNG is too strong relative to your game mechanics, it will not deliver game feel particularly well and players won't be able to sense the mechanics. This creates a Catch-22, where the RPG trope is that the player character is not the player and the difference is (usually) abstracted out with dice, but the player can't feel a thing of what the player character is experiencing through the dice. This is the root reason I dropped rolls to cast spells from my game. A shield spell where you roll 2d4 and target's DR increases by the result feels slow and sloppy because the effect of casting the spell is being squeezed through the hose of rolling those dice. "Pay 2 Mana: target's DR increases by 6," feels sleek, sharp, and crisp in comparison. So if you actually want a mechanic to deliver game feel to the players, I suggest you think long and hard about if you really want to pass it through RNG at all. 

Most features of core mechanics are majority placebo. That isn't to say that they deliver no game feel, but players tend to think themselves into sensing things in the mechanics which aren't actually there in the math. A common thing I've heard experienced D&D players say is that the advantage die which didn't count is "what you would have rolled." No, it's not. The first die you rolled is the one you would have rolled, so odds are 50-50 the higher roll would have actually been what you would have rolled. At the end of the day, the player knows that rolling 2d20 and keeping the higher improves their roll, and that expectation is what creates game feel like you're creating a dry martini by whispering "vermouth" over a glass of gin rather than actually adding any.

For some players that's all that's needed and for others...not so much.

In general, mechanics where players do things have much greater game feel potential than mechanics where players passively allow the mechanics to play themselves. Rolling 1d20 and adding a modifier is boring. It's somewhat better to roll 2d20 and choose the higher result. But it is far better to roll 5 successes, mark 2 off to hit your target, add 1 to deal extra damage, and then spend the last 2 to stun your target. The player is least active in the first example and just executing the actions required to make the game run. The player is most active in the last example, with the inevitable tradeoff being that the last one takes a whole lot of gameplay time.

Because of that, I am cautiously optimistic about your game having stamina affecting the number of dice in a roll. I suggest that you need a physical token of some sort (a poker chip?) to mark stamina penalties because the big issue will probably be remembering your current Stamina status. Rapidly twitching variables are something you must intentionally design around.

The thing which puts me off is you suggesting that you are using a variable target number. If by some chance you mean the success count, ignore this because requiring higher success counts for some actions is a no-brainer.

Dice pools really don't like having the value which determines if a die succeeds or fails change because it disrupts how players view the results. For comparison, my own system uses the full array of step dice, all the way from the d4 to the d20 (being one of only a few games which sensibly includes both in a single core mechanic). Messing with the TN is roughly as disruptive as introducing step dice. Less disruptive in that it's notably faster to vary the TN than fish for step dice, but also more disruptive because the player's viewpoint of the system changes when you change the TN.

Quote from: Eric Diaz on January 11, 2024, 10:18:19 AM
There is nothing wrong with dice pools and I can think of at least two vaguely OSR examples:

- In BECMI, one option for ability checks is rolling 2d6/3d6/4d6/5d6 under ability (which I find a good idea because the numbers work for me).
See here:
https://methodsetmadness.blogspot.com/2016/12/roll-xd6-under-ability-yet-another-d.html

- In the new LotFP draft, it seems that saves are now dice pools (which I find a bad idea for different reasons).

There might be something wrong with some specific implementation, but basically you can get what you're looking for with dice pools: predictability, granularity, dice pools, rare results... just adjust how you use them.

The reason that they don't fit well with OSR (there are a few exceptions as mentioned) is because OSR is meant to be compatible with D&D products that are usually d20.

But if you really want it, it could be done IMO; you could start with using OD&D HD for combat, for example.

Outside of core mechanics such as combat or saving throws, I actually can't see why a skill based task resolution, or some sort of magic or subsystem created via a dice pool would be incompatible?

I'm currently designing an OSR RPG where the core "powers" so to speak are resolved by a dice pool, though it only requires a single success to activate them, and duration/damage/special are determined by how many successes you got.

I also count subsystems within this question.

Quote from: Valatar on January 11, 2024, 04:12:44 PM

Quote from: Steven Mitchell on January 11, 2024, 02:55:10 PM
That's not dice, and a die pool can have similar issues, even if in slightly different places.  That's a failure of GM adjudication, calling for rolls when success should be ruled automatic or impossible, and moving on without touching the mechanics at all.  Or in many cases, a failure of the game to teach GM adjudication in any kind of coherent fashion.

There are very few GM's, even horrible ones, that would call for a roll when a character walked slowly across level ground, with nothing else happening, and no time limit.  Or call for one to keep breathing.  Somehow, that basic fact fails to register as a general principle, though.

I was exaggerating in my example, because of course there shouldn't be rolls to climb out of bed.  But in the d20 system say a level 20 rogue confronted with a DC 10 lock still has that 5% failure chance.  Or, if you go the scaling route of 4e or Pathfinder 2/Starfinder, the level 20 rogue will inexplicably only encounter DC 40 locks at that stage in their career as the entire world has been upgrading their security systems behind the scenes as the rogue has gained levels.

Not to be a pendant, but by the 3e, 4E and 5e rules, natural 1s only automatically fail for attacks and saving throws (similarly except for attacks and saves, a natural 20 is not an automatic success either).

Further, outside of combat, many skills in 3e and 4E allow you to take 10 for the check (5e has "Passive Values" for a smaller list of things and a general guideline that you shouldn't even have a roll unless the DC would be 15+ to begin with) and, if there are no consequences for failure, you can take some extra time to take 20 for the check.

A +9 Open Locks modifier vs. a DC 10 lock will always succeed and there's literally no need to ever roll, even in combat.

Outside of combat, a +0 modifier is sufficient to open a DC 10 lock without even rolling.

With no time pressure, a +0 modifier is sufficient to open an untrapped DC 20 lock without rolling in about 2 minutes.

A reasonable starting Rogue probably starts with somewhere between a +5 and a +8 to their Open Locks check. So outside of combat a DC 15 lock is an action and a DC 25 is two minutes without rolling. The only reason to ever roll in 3e or 4E is if a lock within your take 20 ability is trapped or needs to be opened in less than two minutes.

Edit: and the level 20 4E Rogue won't only encounter DC 40 locks; the rules say those are just the only ones you should even bother rolling for (easier ones automatically get opened and harder ones are impossible).

Quote from: BadApple on January 08, 2024, 05:21:47 PM
The problem I have with dice pools is kind of dumb, tbh.  There's too many things for a player to sort out for a check.

If I've done my job as a GM right, every check has meaning and thereby putting stress on the player.  The more dice a player has to check under stress the longer it takes to sort out and in turn hurts the game's flow and pacing.  I find that three dice is as big as a dice pool gets before it starts to have a negative affect at the table.

I don't understand what you're referring to. There's nothing to "sort out". You roll a number of dice equal to your rank.

My favorite RPGs are mostly dice pool systems.  But, I like ones that most people never heard of. 

D6 Star Wars was already mentioned, but my favorite version is the Mini-Six Bare Bones.  Nuff said. 

Dungeons and Delvers Dice Pool.  First, there is a regular D20 game too, so don't confuse them.  The dice pool version is simplified, but that's a good thing to me.  This dice pool system has stats and skills rated as dice sizes, often with additional dice from specialist abilities. Roll them all, but here is the genius bit.  You only pick the highest 2 dice to be your roll.  Everything is a skill check, including spells.  A fireball spell or a throwing knife are both the same.

Tiny D6 series.  This one is exceptionally simple.  Roll 1D6 for difficult things, 2D6 for normal things, and 3D6 when it's easy.  Get a 5 or a 6 on any dice, you win.  That's it.  This time the percentages are not hard to work out.  33%, 55%, or 70% chances to succeed.  I mostly like it because it covers many genres like superheroes and Mecha pilots.

Quote from: Fheredin on January 11, 2024, 06:10:44 PM

Quote from: Eirikrautha on January 10, 2024, 10:11:46 PM
First, I have considered using two different methods to manipulate the die rolls (and, yes, I've got spreadsheet after spreadsheet from anydice.com to prove it!).  I've considered having anything that gives you a bonus adding a die to your pool.  Anything that makes your task harder will increase the target number (so I'd probably start the target number at 5+ if I went this route).  Mathematically, increasing the target number has a larger effect than increasing the number of dice.  This means the GM can be a little more liberal with bonuses, because anything that adds a penalty will have an outsized effect (so a couple of extra dice won't really offset being prone and blinded, for example).  I know that folks have expressed concerns upthread about not being able to discern the effects of varying multiple features of the die rolls.  Do you think I can get away with this, or would it put you off or cause consternation?

Second, I am structuring the dice pools to average between 3 and 5 dice (d10s), usually with only a couple of bonus dice available.  I don't foresee any rolls being over 7 dice.  But I am looking at mechanics that might play into the number of dice, like losing dice each time you do something strenuous (and regaining dice at a rate each round you don't... sort of a fatigue or stamina mechanic).  Will varying the number of dice each round be too much, especially with varying target numbers?

I'm curious as to your reactions and reasons...

Let me speak as someone with experience tinkering with dice pools which are about as complex as they can get without breaking.

First, about players not being able to discern features; this is a half truth. The root problem is that you are trying to deliver game feel through the metaphorical rubber hose that is RNG. If your RNG is too strong relative to your game mechanics, it will not deliver game feel particularly well and players won't be able to sense the mechanics. This creates a Catch-22, where the RPG trope is that the player character is not the player and the difference is (usually) abstracted out with dice, but the player can't feel a thing of what the player character is experiencing through the dice. This is the root reason I dropped rolls to cast spells from my game. A shield spell where you roll 2d4 and target's DR increases by the result feels slow and sloppy because the effect of casting the spell is being squeezed through the hose of rolling those dice. "Pay 2 Mana: target's DR increases by 6," feels sleek, sharp, and crisp in comparison. So if you actually want a mechanic to deliver game feel to the players, I suggest you think long and hard about if you really want to pass it through RNG at all. 

Most features of core mechanics are majority placebo. That isn't to say that they deliver no game feel, but players tend to think themselves into sensing things in the mechanics which aren't actually there in the math. A common thing I've heard experienced D&D players say is that the advantage die which didn't count is "what you would have rolled." No, it's not. The first die you rolled is the one you would have rolled, so odds are 50-50 the higher roll would have actually been what you would have rolled. At the end of the day, the player knows that rolling 2d20 and keeping the higher improves their roll, and that expectation is what creates game feel like you're creating a dry martini by whispering "vermouth" over a glass of gin rather than actually adding any.

For some players that's all that's needed and for others...not so much.

In general, mechanics where players do things have much greater game feel potential than mechanics where players passively allow the mechanics to play themselves. Rolling 1d20 and adding a modifier is boring. It's somewhat better to roll 2d20 and choose the higher result. But it is far better to roll 5 successes, mark 2 off to hit your target, add 1 to deal extra damage, and then spend the last 2 to stun your target. The player is least active in the first example and just executing the actions required to make the game run. The player is most active in the last example, with the inevitable tradeoff being that the last one takes a whole lot of gameplay time.

Because of that, I am cautiously optimistic about your game having stamina affecting the number of dice in a roll. I suggest that you need a physical token of some sort (a poker chip?) to mark stamina penalties because the big issue will probably be remembering your current Stamina status. Rapidly twitching variables are something you must intentionally design around.

The thing which puts me off is you suggesting that you are using a variable target number. If by some chance you mean the success count, ignore this because requiring higher success counts for some actions is a no-brainer.

Dice pools really don't like having the value which determines if a die succeeds or fails change because it disrupts how players view the results. For comparison, my own system uses the full array of step dice, all the way from the d4 to the d20 (being one of only a few games which sensibly includes both in a single core mechanic). Messing with the TN is roughly as disruptive as introducing step dice. Less disruptive in that it's notably faster to vary the TN than fish for step dice, but also more disruptive because the player's viewpoint of the system changes when you change the TN.

Thanks for the feedback.  The more I get the more I'm rethinking variable target numbers.  Maybe I can add something like additional successes necessary to represent more difficult tasks.  I'll have to ponder it.

Quote from: jhkim on January 08, 2024, 06:06:03 PM

Quote from: Stephen Tannhauser on January 08, 2024, 05:26:57 PM
It's less about what's "wrong" with dice pools, and more about whether they work to deliver typical OSR gameplay goals, part of which is usually evoking the feeling of the old school gaming experience.

I find it weird that stuff from the 1980s like dice pools should be considered new-fangled and not part of old school. Is there is better terminology to distinguish game design that evokes the feel of 1980s games like Star Wars and Shadowrun?

It was the mid 90's, before I was exposed to the games and rules of the 70's.