Why didn't 5E D&D just switch to dicepools?

[Lago PARANOIA]

I'm not saying that 5E D&D should switch to dicepools. In fact I think it's a very bad idea, since D&D (even 4E) loves huge amounts of power scaling and the lock-out effect where if you're not threatened by a hundred guys of a certain level then you won't be threatened by five hundred of the same guys. A linear or bell-curve RNG works just fine for these assumptions and I'm totally cool with that being how D&D is played for the indefinite future.

However, based on what I've seen from the playtest packets and sample adventure -- especially the bounded accuracy crap -- Mike Mearls and friends really, really should've used dicepools. They want a game in which the threat level of an orc never really gets neemed to zero but they also want a game in which there is a definite difference in power between that of a storm giant and a troll. They want a game in which people still have a chance to fail certain tasks no matter how good they get. And they also want a game in which you can fight 30 kobolds at once at low level without the numbers being a quaint formality. That's not how D&D has been played for the past 15 years almost, but I think it's a defensible paradigm shift.

So... why no dicepools?

[Chamomile]

Does it really have to be stated? Because Mearls and co. are hacks who have no idea which mechanics are needed to achieve what results.

[Laertes]

It's because if they did, then codeGlaze's signature line would be proven true.

[RadiantPhoenix]

Doing so would have been counter to their goal of "nostalgia"

[hogarth]

Doing so would have been counter to their goal of "nostalgia"

Yup. Their initial goal was to be every version of D&D to every person, so switching to a mechanic that doesn't belong to any version of D&D would have been a weird first step. Of course, they ended up with a weird, mutant version of D&D in the end, so go figure.

[Cyberzombie]

Dice pools suck in games where you expect people to make a bunch of attacks. If you've got one attack per round, then it's okay to toss 3 d20 and count up hits.

If you want to do routines like claw/claw/bite, hand fighters 3 attacks a round, or have fireballs hit entire rooms of people who all need to roll multi-dice per save, then you have problems. You want to be able to resolve a bear's attack rolls in one toss by taking 3d20 and rolling it. You don't want to have to do that three separate times for each attack it has.

Shadowrun has awesome flavor, but the actual core system is terribly slow and inefficient, as is the case with all dice pool systems. They work okay when you're not making many checks, but in a combat heavy game which involves all kinds of rolls per round, they work terrible.

[Ancient History]

Dice pool systems work fairly well with TCGs like Netrunner, where you can have set resources that refresh every turn.

[John Magnum]

Can you do a fixed-TN dicepool system where you never need to know the literal value of any die? It seems like resolving even a large pile of dice could go very fast if instead of rolling a bunch of polyhedra with numerals on the faces, you rolled a bunch of polyhedra that just had red- or blue-colored faces and you count hits by counting how many blues you rolled. But then you can't have stuff like exploding dice. Depending on how important that is to you, you might not want to give that up.

[Prak]

Yes, but I wouldn't do it by colour. What you do is something like the game Zombie Dice, where there are red, yellow and green dice, that each have three different symbols (brains, footprints and blasts) in different amount. In an RPG you use this to represent skill by giving more skilled characters more dice with successes on them.

So you could do something like everyone starts with ability+skill d20s each with 15 Red and 5 Blue faces. Red is failure, Blue is success. So you basic mechanic is "check colours. More blues = success." Then as people level up they get dice with, say, 12 Red 8 Blue, 10 and 10, 8 Red and 12 Blue, etc.

[Koumei]

That would require actually changing too much for their liking - they would need to rework a lot of mechanics so that everything ran smoothly, and they don't understand a lot of that stuff. Granted, it doesn't look like it runs smoothly now, and I doubt they understand how d20 works, so there you go.

[OgreBattle]

They basically created the d20 system. When you show someone a d20 it's a physical representation of D&D. Take that away and D&D will physically feel 'different'.

[codeGlaze]

It's because if they did, then codeGlaze's signature line would be proven true.

There's still time!
There's splats and .5 editions yet to come!

[Prak]

And Hacker'sDungeon Master's Guide sidebars!

[CCarter]

The core idea of bounded accuracy seems like you keep your same chance to hit forever, but get more hit points and do more damage. That depends on the game's damage-resolution system - the ever increasing pool of dice based on sneak attack or martial bonus dice or caster level - to work, not its task-resolution system.

I think the question is, could a dice pool system do damage in a way that supports that? Or, would you end up either rolling ridiculously large damage dice pools, or with low-power monsters not doing enough damage to get through 'soak' and so being obsolete because they're off the RNG for damage, even though they're on the RNG for attack rolls?

[Username17]

The core idea of bounded accuracy seems like you keep your same chance to hit forever, but get more hit points and do more damage. That depends on the game's damage-resolution system - the ever increasing pool of dice based on sneak attack or martial bonus dice or caster level - to work, not its task-resolution system.

No. It's not fixed accuracy, it's bounded accuracy. The goal is to continue giving out bonuses, but have the attacks stay on the RNG the whole time. The entire point is not that you never get a bigger magic sword or never learn a buff spell that makes you harder to hit, but that no matter how powerful you get you still have a chance of missing Orcs and Orcs still have a chance of hitting you, and there will be some number of Orcs that is scary to you, no matter what level you are or how many bonuses you obtain.

Now, Mearls and company insist that this state of affairs is something that they can achieve by the expedient of simply handing out bonuses slower and slower as characters go up in level and never actually get to the end of the RNG. Now, leaving aside whether this is a goal worth pursuing in the first place, that's obviously a stupid way to go about it. If you used a curved or logarithmic RNG, you'd get there automatically. There are lots of RNGs which simply have "tails" and you could just use one of them and get an Achilles chasing the Tortoise effect without any special discipline on the part of your bonus assignments at all.

-Username17

[Cyberzombie]

If you used a curved or logarithmic RNG, you'd get there automatically. There are lots of RNGs which simply have "tails" and you could just use one of them and get an Achilles chasing the Tortoise effect without any special discipline on the part of your bonus assignments at all.

How would you achieve that? It seems like you'd need some kind of reverse bell curve, but I have no idea how you'd get that with dice.

[Lokathor]

Well, not to put too fine a point on it, but you can have a chance to always miss with... dice pools. And similarly, your can always potentially roll as many hits as you have dice in your pool, even though that's exceedingly rare once you've got just a few dice.

In terms of being able to scale Damage / Soak up and down the same as Attack / AC are going up and down, well one of those two pairs should scale up and down, and the other probably shouldn't. Otherwise you get really complicated computations of what the average results are gonna be because you've got four shifting variables.

[RadiantPhoenix]

If your goal is to ensure you always need to consider the orcs, the defender should roll -- a 100% chance of hitting isn't that different from a 99% chance of hitting, but a 0% chance of hitting is worlds away from a 1% chance of hitting.

Also: "Achilles chasing the tortoise"?

[Username17]

If you used a curved or logarithmic RNG, you'd get there automatically. There are lots of RNGs which simply have "tails" and you could just use one of them and get an Achilles chasing the Tortoise effect without any special discipline on the part of your bonus assignments at all.

How would you achieve that? It seems like you'd need some kind of reverse bell curve, but I have no idea how you'd get that with dice.

As Lokathor points out, Dicepools simply do that. A dicepool is an exponential function, but the tail remains. No matter how many dice you roll, there is still a chance of not rolling any hits. The dicepool's tail is actually infinitely long if need be, there is literally no number of dice you could possibly add that would fail to live up to the stated demands of bounded accuracy.

A regular curved roll (rolling XdY and adding the results to together, where X > 1) also has a tail. It's not infinitely long, but it can be very long if need be. If your RNG was 4d10, there are five numbers on either end of the curve which collectively come up less than 1% of the time. Now obviously there is an end to that RNG, but you'll stop noticing or caring about marginal improvements long before they literally stop being able to be added. That same +1 bonus that shifts you more than 6% in the middle of the curve shifts you just one percent of one percent when you're near the edge. One could rather easily make the claim that you were going to stop handing out bonuses before you literally got to the end of such an RNG, and that would fulfill the bounded accuracy claims as well.

-Username17

[Cyberzombie]

As Lokathor points out, Dicepools simply do that. A dicepool is an exponential function, but the tail remains. No matter how many dice you roll, there is still a chance of not rolling any hits. The dicepool's tail is actually infinitely long if need be, there is literally no number of dice you could possibly add that would fail to live up to the stated demands of bounded accuracy.

I always thought the goal of bounded accuracy to make sure that orcs are always relevant, not just to always give them a minute chance of hitting, since we already have that with the natural 1/20 rule.

Wouldn't it be easier just to make natural 19+ autohit if the goal is purely to prevent defenses from totally shutting out weak attackers?

[schpeelah]

How would you achieve that? It seems like you'd need some kind of reverse bell curve, but I have no idea how you'd get that with dice.

On a curve a +1 is big in the middle but small once you've accumulated 5 of them. As opposed to a flat roll where you hand out a +4 as the first bonus and make later bonuses smaller for the same effect.

Also: "Achilles chasing the tortoise"?

One of Zeno's paradoxes: Achilles is chasing a tortoise who has a 100m head start. When he reaches the point the tortoise was at the start, it'll have moved 10m. When he reaches that point, it'll have moved 1m etc.

[Cyberzombie]

How would you achieve that? It seems like you'd need some kind of reverse bell curve, but I have no idea how you'd get that with dice.

On a curve a +1 is big in the middle but small once you've accumulated 5 of them. As opposed to a flat roll where you hand out a +4 as the first bonus and make later bonuses smaller for the same effect.

On a linear RNG, at worst, your chances get halved by a +1 modifier, that's where you need a 20 instead of a 19. If you're talking about the difference between rolling 18 on 3d6 versus rolling a 17, then your odds are becoming worse than half. With a curve it's actually easier to push someone to irrelevancy. While bonuses do diminish in value at the ends, by that point it's because you're generally talking about so low odds that you're worse off than using a linear RNG like a d20.

If you wanted to keep people relevant, you'd probably have to do something weird like redefine the faces of a d20. Keep it linear but say for instance a 12 counts as a 14, a 14 counts as an 18 and a 16 counts as a 25, or something like that where it takes a truly huge bonus to totally push someone off the upper ends of the RNG. It'd be stupidly complicated though since you'd have to constantly be looking up an arbitrary table with every roll. And it'd probably be easier just to say that 19+ or 18+ always hits.

[Mord]

Not to disagree with any of the statements above, but I think it's worth considering the flip side of the bounded accuracy coin. Namely, that the concept theoretically allows low-level PCs to engage with higher-CR antagonists than would be possible otherwise. I can't imagine anyone giving a real shit about level 15 PCs fighting a bottomless horde of Orcs as compared to, say, a party level 5 PCs fighting a single Chromatic Dragon in some kind of climactic set piece battle.

It works out to basically the same thing, but I think the latter use works out to be more fun for the people at the table when you consider the sheer number of bullshit rolls involved in running a combat between 5 PCs and 25 Orcs... Unless the party wizard just annihilates them all with a cloudkill or something in round 1, which kind of puts the kibosh on the idea for a totally different reason.

[Prak]

The problem is that "always has a chance to hit" is not the same as "a chance to win." As Cyberzombie pointed out, we already have a set up where low level creatures always have a minute chance of hitting the dragon. However, if you throw a party of level 5 PCs against a giant dragon, the dragon is very likely to win just by hitting more frequently and harder. If the PCs had like 20 other guys with them, then they may have a better chance of doing significant damage to the dragon, but a lot of the guys are going to die, and without special plot armour rules, there's no guarantee that it won't be the PCs who die.

[OgreBattle]

Would Warhammer be an example of bounded accuracy in action? There's always a chance for your Bloodthirster to get stabbed by a puny human.