It's really about the combination of the dice and the DC (and the roll bonus).

With a D20: If I have, say, +2 in some skill, and the DC is 14, then I need to roll an 12 or higher to succeed. My chance of success is 45%. If I got a +1 from somewhere, then it's 50%, and an additional +5% for each +1. That's the flat curve.

If, instead, I'm using 3d6, my chances start at 37.5% (iirc). +1 shifts that to 50%. +2 shifts it to 62.5% (so those first 2 are worth +12.5% each). +3 brings it to a little over 74%. +4 to a little less than 84%. As you can see, the returns start diminishing --- the difference between +7 and +8 is just over +1%, and the difference between +8 and +9 is about 0.5%...

In other words, if you were betting on dice rolls, your bets would (well, should, but there's no accounting for being bad at gambling...) change between d20 and 3d6. D20 is closer to roulette, while 3d6 is closer to craps (please excuse the grand oversimplification). When you're playing a tabletop, at least if you're thinking in game-like terms of success and failure, you are indeed gambling on dice rolls. The variances matter --- different game feel from different math. D20 is more chaotic. 3d6 is more predictable --- player choices matter a tad more, both because tiny +1s matter more when you're close to the median, and because you have more reliability in your dice...

Quote from kenclary >>

It's really about the combination of the dice and the DC (and the roll bonus).

With a D20: If I have, say, +2 in some skill, and the DC is 14, then I need to roll an 12 or higher to succeed. My chance of success is 45%. If I got a +1 from somewhere, then it's 50%, and an additional +5% for each +1. That's the flat curve.

If, instead, I'm using 3d6, my chances start at 37.5% (iirc). +1 shifts that to 50%. +2 shifts it to 62.5% (so those first 2 are worth +12.5% each). +3 brings it to a little over 74%. +4 to a little less than 84%. As you can see, the returns start diminishing --- the difference between +7 and +8 is just over +1%, and the difference between +8 and +9 is about 0.5%...

In other words, if you were betting on dice rolls, your bets would (well, should, but there's no accounting for being bad at gambling...) change between d20 and 3d6. D20 is closer to roulette, while 3d6 is closer to craps (please excuse the grand oversimplification). When you're playing a tabletop, at least if you're thinking in game-like terms of success and failure, you are indeed gambling on dice rolls. The variances matter --- different game feel from different math. D20 is more chaotic. 3d6 is more predictable --- player choices matter a tad more, both because tiny +1s matter more when you're close to the median, and because you have more reliability in your dice...

It's a chaotic world

The DM can set a DC, but the die roll is the thing that accounts for a bit of rust in the lock mechanism, or the fact that the person you're trying to persuade is having a bad day and is grumpy, or the sudden noise that distracted your mark as you pick their pocket, or the fact that your grandfather loved telling stories about the period of history you need information on

Quote from MrRezister >>

Let's start a fight:
2d10 🎲 makes more sense than 1d20 - change my mind.

Every time I roll a d20, I am just as likely (percentage-wise) to end up with a critical failure as I am to just barely succeed, because every face on the die (theoretically) is just as likely to come up as any other.

Whereas with 2d10, the likelihood of rolling an "average" check or save (10, 11, 12) is significantly higher than getting a critical failure (double 1's) or critical success (double 10's). Not only does the 2d10 model more accurately represent a realistic potential outcome, but it also makes your skill bonuses (or penalty) more meaningful because it reliably increases (or decreases) the likely average of all your rolls, instead of just sliding the window one direction or the other.

What do you think?

Or you could roll 3d20 at all times.  Define a normal roll as taking the median die, advantage as the greatest die, and disadvantage as the lowest die.

Quote from Yurei1453 >>

What Saga and I are suggesting is that examining single rolls, in isolation, is disingenuous because the game experience isn't about one single roll made in isolation, it's about the cumulative experience of thousands of rolls across an entire campaign.  When those rolls are considered as an aggregate whole, d20 resolution is swingy and obnoxious and results far too often in aberrant events that defy reason.  The rogue breaking her thieves' tools on a DC5 lock, the wizard failing an Arcana check only for the barbarian to pass the same check and look at the wizard all "are you stupid or something?", the plate-armored fighter not managing to avoid a single enemy blow...there's lots of ways d20 resolution produces aberrant results.  It's so common it stops being fun.

1. Yes, I agree that Pangurjan is missing your point - d20 does miss your point, it does have more edge cases. You are more like to roll a 19 or 20, and likewise a 1 or 2.
2. It's your assertion that these defy all reason and is obnoxious. This is your very subjective view. As others have pointed out, they find it better than an xdy system.
3. I have a simple solution for you if you don't want your master thief to break their tools breaking into a kid's piggy bank - play by the rules. It's as simple as that. You've been told this a lot of times, you can't complain about a rules making you do something that the rules says can't happen, now drop it.
4. A relative +5 on a skill makes you win around 75% of contests. I think that's fair for a mental ability check. Having fresh eyes, being free of preconceptions, just plain old brain farts, it's not uncommon for less intelligent people to make a connection that the intelligrnt person missed. Where the real gap comes is in specialised areas, the equivalent being things like spellcasting - which really is beyond a dumb Barbarian anyway.
5. Plate gives you an AC of 18. Against someone with a to hit of +5, that's a 65% of an attack missing, or 35% chance to hit. Let's say that that they're targeted by four attacks, that's...1.5% that they'll be hit every time. That's roughly one in 70 combats - if you have 3 combats per level, being targeted 4 times per combat by an attack that compares your AC, and your campaign runs to L20, that's an expected occurrence of once per campaign (you may be targeted more often per combats by an AC- relevant attack, but that also makes it much less likely to meet your criteria for being hit every time). If that's happening frequently, you seriously need to invest in a dice tower or new dice or something, because the rolls aren't fair. Regardless, 2d10 gives the same proportion of hits at that point (entirely incidental, I didn't look it up until after). 3d6 is lower, at 25.79%, but as soon as that attacker gets a +6 to hit, it's about as likely to hit as a d20 (37.5%v40%).  That happens fairly early on in my experience (monsters seem to get itnot long after characters generally get plate anyway), and +7 (again, I'm level 9 and facing this level of attacks) are slightly less likely with a d20 so moving to 2d10 or 3d6 isn't a solution to your problem.
6. I think your problem is more with the value of the modifiers vs the value of rolls than the probability distribution. The value of the roll eclipses the value of the modifier. That's a 5e problem rather than a d20 problem.

Quote from Linklite >>

Quote from Yurei1453 >>

What Saga and I are suggesting is that examining single rolls, in isolation, is disingenuous because the game experience isn't about one single roll made in isolation, it's about the cumulative experience of thousands of rolls across an entire campaign.  When those rolls are considered as an aggregate whole, d20 resolution is swingy and obnoxious and results far too often in aberrant events that defy reason.  The rogue breaking her thieves' tools on a DC5 lock, the wizard failing an Arcana check only for the barbarian to pass the same check and look at the wizard all "are you stupid or something?", the plate-armored fighter not managing to avoid a single enemy blow...there's lots of ways d20 resolution produces aberrant results.  It's so common it stops being fun.

1. Yes, I agree that Pangurjan is missing your point - d20 does miss your point, it does have more edge cases. You are more like to roll a 19 or 20, and likewise a 1 or 2.

The thing is that 1s and 2s aren't edge cases, and neither are 19s and 20s. Without crits, there's nothing special about them. I'm not missing the point, I'm dismissing it. I'm dismissing it because the "you're going to get too many unexpected results across thousands of rolls" argument is IMO bogus. You might roll thousands of skill checks over 20 levels of playing a character, but very few of them will be edge cases. Setting aside the fact that making a trained rogue roll to pick a DC 5 lock is inane, unless we're talking external modifiers so the DC isn't really 5 there's exactly one possibility of this failing - a Dex (or Int or whatever the DM chooses to call for) modifier of +1 and a proficiency modifier no higher than +2. Nothing else fits the description of the example. Not being proficient means no training, not even a +1 from attribute means no talent or the check being out of their wheelhouse, and a total bonus of at least +4 means no failure. So we're talking about a DC 5 lock check, which should be rare enough to begin with, having to be performed by a character with a proficiency and attribute combo that's probably just as rare, and in the already extremely rare case both of these happen concurrently the actual chances of failure are 5% assuming no advantage can be provided and no other bonus can be given. The chances of all this lining up are astronomical. Of those thousands of skill checks maybe one or two percent will be such an edge case setup - let's say 10,000 rolls and 5% having the potential to be weird like that. Then we end up with 500 of these rolls, 25 of which will result in failure due to rolling a 1. Over 20 levels, 25 individual checks. And they won't all be DC 5 locks or Barbarians getting sciencey, they'll likely be for 25 completely different things. That's a non-issue in my book. As you suggest, just play by the rules and it's fine. There will not be weird strings of "this shouldn't happen" failures (or successes) on skill checks because those checks will be rare to begin with, and by the rules should even more rarely being called for. Having critical effects on skill checks isn't in the rules at all, so that doesn't even warrant addressing.   

I think the observation that using a system that emphasized average results (low standard deviation) favors the more powerful side is very astute compared to the alternative of a system that allows equal probability of results over the full range (high standard deviation).

The journey will have more excitement using the d20.  That sounds positive to me.

Quote from MusicScout >>

I think the observation that using a system that emphasized average results (low standard deviation) favors the more powerful side is very astute compared to the alternative of a system that allows equal probability of results over the full range (high standard deviation).

The journey will have more excitement using the d20.  That sounds positive to me.

 To be honest, the increased rate of freak results is actually quite low. I showed the working for plate armour not tanking any shots in a combat - it happens maybe once  a campaign. I just worked out how often a +5 would fail a DC15 check but a +0 would succeed for a d20 roll versus 3d6. It's 12.5% v slightly over 6% respectively. Only in 6% of rolls would the aberrant result be eliminates. 1 in 16 rolls. That's...not very often. Usually, it's not +5 spread either, which reduces that margin. It's just not worth reducing crits from 1 in 20 to 1 in 100, plus the rest of the other points I Iisted and what the others have raised.

There is an argument for the 2d20 that we use in Star Trek Adventures and other systems. There's a lot of flexibility there and allows armour to be actively used (ie rather than just increasing the odds that an attack will miss, but actually decreases the damage taken on a hit, etc). But then, that's a view of the engine itself, rather than how many dice and what dice you use in 5e.

Using GURPS as an example because it's the dice pool system I know the best...that isn't the case at all, Arnt.

Moving target points drastically shift the odds, and the GM applying bonuses or penalties to the roll - or the player taking bonuses or penalties as part of certain abilities - can make having a very high score worthwhile.  In GURPS you roll against your target skill, and having a skill rating of 16 means you can only flub any given skill check on a critical failure.  Hell, even just a rating of 14 in a skill can make you mostly immune to "ordinary" failure.  And yet the one game I got to play, there were multiple characters with 20+ skills, and one with a whole-ass 25 in their Broadsword skill.  Why?

Because you don't get unmodified rolls hardly ever, there's always some penalties associated with the given situation.  The master swordswoman with an enormous Broadsword number could fight with severe handicaps, or attempt combat maneuvers that normally come with a severe to-hit penalty relatively safely.  She was more than rewarded for her investment, while my own character with a huge slew of barely-trained skills powered by her freakishly high DX was far less able to cope with the realities of adventuring.  Her odds of pulling off any of her plethora of talents fell off sharply once reality intruded; she could take a stab at a lot of things but was nowhere near as good at any of them as a true specialist.

Switching D&D to a dice pool would involve reworking the whole resolution engine, yes.  It's not going to happen.  But it's absolutely possible to make a dice pool resolution system that values and rewards skill/training, just like it's possible to make a d20 resolution system that doesn't give a single fat frog fart about skill/training D&D 5e.  Simply depends on your design chops and what you value in your game.

It is all artificial. The better the PCs, the harder the DM makes the game, plain and simple. Just like those video games (like Cyberpunk ‘77) where you keep getting slightly better versions of the same weapons, but the enemies also get more health bar…. It’s a fugazzi, but it’s fun and it feels cool.

Higher modifiers will give more options. DC 20 to scale a cliff so you can approach the warlord's stronghold unnoticed is not an option with a +3 Athletics modifier, but if it's +13 it's something you can at least consider. Sneaking past alert guards can be tried with a +12 or more in Stealth, but should probably be avoided if that modifier's +4 or less. Deceiving the vizir who's preventing you from seeing the caliph is not a good idea with a +2 Deception bonus, but with a +14 you can give it a try. 

Of course, all of the above can easily be circumvented with magic once you're past tier 1 play, but that's an issue for another thread. 

Quote from ArntItheBest >>

Quote from kenclary >>

<snip>

This analysis means that it is pointless to get expertise and a high modifier in any given skill. Because you have the diminishing returns, now someone with a high strength is just as likely to make a skill check as someone with expertise in Athletics. In an xdy system, to really have a good character, you would put your proficiency and expertise in skills that you have a low ability score for.

To be super clear, I don't think 3d6 could just be "dropped in" as a substitute for d20 and the game would keep on working the same (that was closer to true in 3.Xe, but 5e is more committed to d20). To get the most out of it / to maximize the fun, I'd want to redo large swathes of the game, especially including the kinds of bonuses and penalties a roll can get. It would be a major new system to the game, one that essentially replaces most of the math. In my experience, such a system would still have lots of value for high skill / expertise; it would just work differently.

I personally understand that that sort of thing isn't worth the effort for many people. Or, perhaps, boils down to "just play a different game." That's fine.

The d20 is very tied to D&D's brand, so I doubt WotC would put real work into removing it.

Quote from shoak111 >>

All this talk about chaos vs. reliability misses the point, because switching to a 2d10 system is not practical.  It would require a complete overhaul of almost every stat and ability in the game if you want to maintain balance.

Basically, with 2d10, you get diminishing returns on bonuses at the more extreme disparities between bonus and DC - whereas in a d20 system, each bonus counts the same.  So effective min-maxing in a 2d10 system is more about trying to be average in many different things, rather than being at extremes.  So the term "min-maxing" in a 2d10 system becomes somewhat of an oxymoron.  Not such a big deal at first glance.  But delve deeper and you will find dramatically different character builds, from feat choices, to stat scores, to character classes, etc.

But the impact on PC choices is only a small part of the overall changes on the game that would be wrought from a switch to 2d10.  Of course, creatures don't get choices in their stats, so they would be stuck with stats useful in a d20 system (and a CR based on those stats) that are far less effective in a 2d10 system.   Level appropriate enemies who aren't proficient in a spells save type would typically have half or less of their current chances to save if you switched to 2d10 (instead of a 20% chance of rolling a 17+ in the current d20 system, they are stuck w/only a 10% chance in a 2d10 system).  So that would lead to different choices in which spells you choose - those save or die spells are looking even yummier now maybe.

And forget about having lower CR enemies for the PCs - those orcs' chance of getting a 19-20 to hit just went down from 10% in d20 to a paltry 3% in 2d10.  So now maybe you need to throw 60 orcs against the PCs instead of 20......

Don't even get me started on how much it affects the ADV/DIS mechanism....

These are but a few of MANY examples of problems with shifting to a 2d10 system.  The whole game is balanced around a d20 system.  Changing that fundamental aspect would force a DM to change almost everything else in the game in order to keep it balanced.  

 

 

 No it doesn't.  Changing to a system like 3d6 would require that kind of change, but 2d10 is not so significant.  

No it doesn't!  If one were to use a 2d10 system, that's not going to make them suddenly want to take 6 13's in attributes rather than 8, 10, 12, 13, 14, 16 or spread out more proficiencies rather than take expertise.  Min-maxing is about being the best you can be at a certain thing you value more than other aspects of play.  Peoples problem with d20 is that is their "best-ness" is easily overruled/negated by sheer dumb luck.  Reducing the influence of sheer dumb luck over the final outcome, does at least make it feel as though the "Best-ness" that was invested in is always noticeably present.  Again the diminishing returns at the upper end are at the very least a price those people are willing to pay for likewise diminishing the chances of a lower roll; and at best, those diminishing returns can be compensated for by slight alterations to the critical mechanic to maintain the 5% chance there.  

I'm not sure if your third paragraph is describing favorable or unfavorable effects re:fun, but again, you don't need to assume that you need to stick to one dice system for everything in the whole game.  If it turns out that d20 is better for combat, but 2d10 is better for social encounters for example, you can adjudicate on the fly which dice you ask for when calling for a roll.  I don't think many of the proponents of integrating some 2d10 into things are folks who actually care about the occasional mishap in a fight as much as they do when their Bard of Planes-spanning fame, renowned for bringing a tear to Nerull's eye when reciting the epic tale of Orpheous, etc.  Rolls even a 2 in Tiny Thorpe Tavern, and get's booed and rotten-vegetable-d buy level 1 peasants.  -  I say a 2, but I'm going to agree with Yurei here and say 'RAW-be-damned: every game I know of too does indeed use fumble/crit mechanics for everything and not just combats - even ME!; hence I really don't mind providing players who are really bothered by it, an alternative to having to put up with it.

Yeah, well, orcs are people now apparently and so shall ever after be slotted into the team hero side of the board rather than used as disposable low CR encounters - along with kobolds and goblinoids.  So if the objection is I'll need to find a more challenging monster to replace orcs with, in order to balance an encounter against a party that no longer rolls low; well, I need to find new disposable npc's to replace orcs with anyway, so I might as well step up the CR by one notch.

and that's also a 5e thing.  I'm going to assume in this case that whoever said it above is right, and ADV/DIS was WoTC's attempt to add in some bell-curve effects without deviating from the 'sacred-cow'.  Now that they've apparently decided to deviate from said metaphore in several other areas of the game, I don't think it's too much of a problem to suggest to them for 5.5/6 that perhaps Advantage could mean using 2d10 for that particular check vs. 1d20 for non-advantaged checks. 

Does anyone have a graph than compares 2d20-lowest to 2d10?  or that shows what 2d10 would be like under the 3e crit mechanics prior to bounded accuracy?

Quote from kenclary >>

The d20 is very tied to D&D's brand, so I doubt WotC would put real work into removing it.

Yeah, I can't think of too many logos that incorporate 3d6

Since there is no critical failure/success of ability checks anything with a DC of 10 (let alone 5 which can often be auto success at level 1) will be auto success with a +10 modifier in the current D20 system.

Thieves tools don't break. That barbarian with -2 arcana doesn't automatically get the info on a crit. If you feel the barb shouldn't be able to roll on the check the DM should set the DC to 19 and the barbarian can't succeed but the wizard can. I find a >+10 generally makes it so you don't have to worry about ordinary rolls in DND and they just don't get rolled. Characters with expertise are rewarded heavily in DND and often can ignore ordinary checks as well. 

A lot of the complaints seem to revolve around homebrew rules for crits/crit fails. It seems it would be much more simple in this system to remove the homebrew rules causing the frustration. Then to rework a system so those crits/crit fails happen less.

After thinking about it my argument for keeping it linear vs a curve is: Setting a DC based on what % chance of success I would like someone to have becomes much more difficult using a dice pool than a single die. I'd have to either do the more difficult math on the fly or have a chart beside me for all the percentage chances. I guess eventually I'd get a feel for it but a lot of work until then.

You could also switch to a D100 to lessen the crits/crit fails while retaining the linear and easier understood % chance of success for setting DCs. Or homebrew in a rule that overrides your other homebrew where it's only a true crit fail if you roll a d100 after and also get a 1 on that and same for crits - this would reduce the chance of crit fails/successes to 0.05%. Maybe that's too low and you prefer a 2nd d20 instead for a 0.25% chance instead or a d6 for 0.8% chance while retaining a simple linear DC system for setting difficulties. A d10 to confirm crits/fails would get you to near the same chance as 3d6 to roll a true crit fail (0.5% vs 0.463%~).

There seems to be lots of ways to eliminate that portion of the problem without changing the way everything else is resolved or require stat block changes. This would still require we look at abilties/items that are affecting crit chance specifically unless you only applied the extra roll to crit fails. This feels like it's just fixing homebrew with more homebrew though.

Quote from MusicScout >>

I think the observation that using a system that emphasized average results (low standard deviation) favors the more powerful side is very astute compared to the alternative of a system that allows equal probability of results over the full range (high standard deviation).

The journey will have more excitement using the d20.  That sounds positive to me.

 It will be more exciting, yes, but whether or not that is positive is again a matter of personal preference, I think.  It's like how some people love going on the rollercoasters at an amusement park while others wait for them at the hot-dog stand or go to like the merry-go-round instead.  Some people like a certain amount of excitement, but then more than that is too much. -shruggs 

The big difference in preference I think is that the bell curve adds a touch more realism and so would for example be good for low-fantasy settings versus the linear for high-fantasy settings.  There are enough campaign worlds to around for each to have their own, but still, I'd also like to see a more integrated system that can work in a wider array of settings for groups accommodating multiple playstyles and preferences.

This is an year old post, reignited three days ago in a discussion that is endless. All that facts laid out are simple:

- Dice pools are naturally more bound to average than a single dice roll. That is the only absolute truth in any statement.

Is it better? Well, really is a matter of preference sprinkled with personal experiences. Players who had adversarial DMs will probably be drawn towards a dice pool system, but, honestly, if you're playing with one of these guys, he/she will just come up with other ways to make your day shitty.

If we go back to the essential question, does a dice pool makes for a better D&D game? Maybe, but only if you have the right people playing with you.

Despite all the mechanics involved, that's what D&D is about and that's what makes a game good or bad. Dice pools or single dice are such a minor thing when compared to that. I really question if it is worth all this energy put into a discussion.

Quote from ArntItheBest >>

I think the randomness actually adds to the realism. Even the best, most surefooted acrobat can step on a seemingly sturdy stone to find that it is loose, or slip on a patch of black ice that they just didn't see. No one is ever successful 100% of the time no matter how good they are at something.

The 5% is the unrealistic thing; not that the occasion should ever happen at all.  A curve doesn't give you 100% success; it just shifts your rate of success up for easy tasks.

The solution is to buy a bag of dice, and conduct the "saltwater" test until you find a d20 that rolls high most of the time.

https://www.youtube.com/watch?v=_HhFz7fsFKk

I agree completely with the 2d10 as that creates a bell curve.

Quote from KanedaSyndrome >>

I agree completely with the 2d10 as that creates a bell curve.

So does an actual bell when it casts a shadow.