The Gaming Den

What is your experience with A/D?

It adds a bell curve to roll results (at least as far as I can tell, I'm not sure) without breaking RNG.

IMO it's a good thing.

Frank liked it so much he made it one of the core features of his Fighter remake.

In isolation it is a good thing. Since the D&Dnext developers are functionally unable to keep stacking bullshit bonuses away from class abilities and spells, it is broken as shit. Put race/class/equipment combinations together correctly and the RNG never matters. Especially since monster numbers are static and designed to average around +4.5 or 14.5 (for attack/defense)

Plus they hand out A/D like candy: the majority of classes can hand out disadvantage innately, and there are a bunch of spells that do one or the other for entire fights.

Rolling two d20s and picking the better is worth, on average, about +3. However, you can normally only get three separate +3 bonuses added together before you're off the RNG altogether. The reroll doesn't break the RNG the way a fourth +3 bonus would.

If you insist on using a d20, and you insist on adding bonuses together, and you don't want to break the RNG, then adding a reroll in as one of your bonus types just makes good sense. But it's not a panacea. While you could add more rerolls, that gets really time consuming and annoying really fast and results start clustering pretty fast (lust look at what a clusterfuck Silhouette is when dicepools get moderately large). Practically speaking, one reroll is probably all you're going to want to allow, although it's not the end of the world if you allow 2.

Mike Mearls fapping to how it solves everything is just bananas. It doesn't solve everything. It solves the problem of adding a fourth +3 bonus without breaking the RNG. But it doesn't solve the problem of adding a fifth. Basically, it lets you go about one more standard deviation than a flat RNG is normally able to accommodate. And it does it by making your RNG no longer flat, so not much of a surprise there. Switching to a 3d6 system would do much the same to your ability to model probabilities.

-Username17

Since they hand out A/D like candy and since further books down the line must introduce power-creep or they won't sell, we can surely expect some kind of DOUBLE or TRIPLE ADVANTAGE in a couple of years.

And since they'll need to build boss monsters that can somehow survive for 3 or 4 rounds against the onslaught of a party rolling several d20 for each hit or save or lose spell, they'll introduce a number of successes mechanic for hitting the toughest monsters. So once we get to killing Lolth or Asmodeus, their AC will be listed like 15 (2). They'll just assume you'll be using several kinds of advantage, but will require you to confirm the hit.

Advantage will lead to dicepools in D&D. Remember, you read this here first!

Yeah, the problem with re-rolls as a standard bonus is it takes away the ability to resolve multiple attacks at once. In Shadowrun you can't resolve 10 guard dogs attacks simultaneously and in D&D you can and that can be a huge time saver. I think relying on re-rolls to not break the RNG is getting out of a frying pan and into a fire. There are small advantages but negatives that are at least as sizable. I wouldn't support it in general.

deanruel87 wrote:Yeah, the problem with re-rolls as a standard bonus is it takes away the ability to resolve multiple attacks at once.

Yeah, as many issues as there are with A/D, this isn't one. There are lots of ways to generate multiple dice rolls in groups simultaneously, no matter if you use physical dice, web pages or apps.

FrankTrollman wrote:Rolling two d20s and picking the better is worth, on average, about +3.

What's your math on that?

With 2dN (choose the higher), our possible results are 1 through N, N=20 in this case.

The probability (P) of getting 1 is 1/20^2 (only one way, for both d20s to get a natural 1)
P(2) = 3/20^2 (only three combinations permit this to happen)
P(3) = 5/20^2
P(k) = (2k-1)/20^2
P(20) = 39/20^2

For our purposes, the average is
* sum of k=1 to k=n of k(2k-1)/n^2
* distribute the 1/n^2 and focus on the sum of k(2k-1)
* This is the sum of 2k^2-k.

The sum of k, k=1 to k=n, is n(n+1)/2
The sum of k^2, k=1 to k=n, is n(n+1)(2n+1)/6
So the sum of 2k^2-k = 2n(n+1)(2n+1)/6 - n(n+1)/2
= [n(n+1)/2][(4n+2)/3 - 1]
= [n(n+1)/2][(8n-2)/6]
= n(n+1)(8n-2)/12

We return to that 1/n^2 from earlier. Then we have

(n+1)(8n-2)/12n

This is the formula. Since n=20, we run the formula and get

(21)(158)/240=13.825

As a single d20's average roll is 10.5, you can get away with saying that 2d20 (drop the lower) is equivalent to a +3 bonus.

If that's too complicated for you, d20s are actually just small enough that you can total up all the possibilities and divide. There's 1 result that is a 1 (1,1), 3 results that are a 2 (1,2; 2,1; 2,2), 5 results that are a 3 (1,3; 2,3; 3,1; 3,2; 3,3), and so on all the way up to 39 results that are a 20. And of course, four hundred total possible results. For disadvantage it's just the opposite, where there are 39 results that are a 1 and only 1 result that is a 20 (and still 400 total possibilities).

-Username17

Voss wrote:Yeah, as many issues as there are with A/D, this isn't one.

You are wrong on first inspection. No matter how you slice it re-rolls are definitely a time cost if resolving many rolls simultaneously. That's obviously true. Now if your group has 10 sets of d20's in pairings of easily differentiable colours that's great but even in that ideal scenario there is a time cost of gathering and using correctly matched colour pairings of dice for those rolls. Absolutely no matter how you separate things there IS a time cost for re-rolls so it IS magnified when trying to resolve many rolls at once. That's just logically true.

deanruel87 wrote:

Voss wrote:Yeah, as many issues as there are with A/D, this isn't one.

You are wrong on first inspection. No matter how you slice it re-rolls are definitely a time cost if resolving many rolls simultaneously. That's obviously true. Now if your group has 10 sets of d20's in pairings of easily differentiable colours that's great but even in that ideal scenario there is a time cost of gathering and using correctly matched colour pairings of dice for those rolls. Absolutely no matter how you separate things there IS a time cost for re-rolls so it IS magnified when trying to resolve many rolls at once. That's just logically true.

Well, you could probably use the 10 minutes between each of your turns to get your dice ready if you have multiple attacks. You could also do like we do in 40k for twin linking, roll your dice, for any misses or 'bad' outcomes, pick up those dice and re-roll which eliminates both multiple dice and having to match/pair dice to attacks.

I don't think the amount of time used by advantage/disadvantage is really noticeable enough to bother caring.

So to be clear. Your points for why re-rolls don't take extra time when resolving multiple rolls are 1) you can take extra time to deal with the extra time they require, and 2) you can roll them sequentially thus no longer resolving them simultaneously and taking more time.

deanruel87 wrote:So to be clear. Your points for why re-rolls don't take extra time when resolving multiple rolls are 1) you can take extra time to deal with the extra time they require, and 2) you can roll them sequentially thus no longer resolving them simultaneously and taking more time.

Previn wrote:I don't think the amount of time used by advantage/disadvantage is really noticeable enough to bother caring.

I don't see a need to further respond to post so poorly reasoned that I'm left seriously questioning if you even read, let alone understood what I posted.

deanruel87 wrote:

Voss wrote:Yeah, as many issues as there are with A/D, this isn't one.

You are wrong on first inspection. No matter how you slice it re-rolls are definitely a time cost if resolving many rolls simultaneously. That's obviously true. Now if your group has 10 sets of d20's in pairings of easily differentiable colours that's great but even in that ideal scenario there is a time cost of gathering and using correctly matched colour pairings of dice for those rolls. Absolutely no matter how you separate things there IS a time cost for re-rolls so it IS magnified when trying to resolve many rolls at once. That's just logically true.

Oh, good. As long as this turns into someone totting up individual seconds before screaming that <task> TAKES 10 MINUTES!

I don't care about your apparent inability to handle and manage simple and trivial tasks.

Without paired dice, A/D will take noticeably longer for large combats.

Oh No! An edition of D&D is bad at large-scale combats! Stop the presses!

D&D has always sucked at larger combats, so adding a rule that ends up discouraging DMs from running a large number of enemies seems fine to me.

virgil wrote:With 2dN (choose the higher), our possible results are 1 through N, N=20 in this case.

[etc.]

As a single d20's average roll is 10.5, you can get away with saying that 2d20 (drop the lower) is equivalent to a +3 bonus.

It should be noted that your calculation of the mean is assuming that target numbers are uniformly distributed between 1 and 20 (e.g., that rolls that require a 1 or higher to succeed are just as frequent as rolls that require a 10 or higher to succeed and just as frequent as rolls that require a 20 or higher to succeed), which is a big assumption if you're talking about D&D.

It may be obvious, but rerolls are not much help at the upper and lower end of the spectrum (if you need a 20+ or a 1+ to succeed, a reroll is worth about a +1 bonus), and are maximally helpful if you have a 50/50 chance of succeeding (if you need an 11+ to succeed, a reroll is worth about a +5 bonus).

TiaC wrote:Without paired dice, A/D will take noticeably longer for large combats.

Oh No! An edition of D&D is bad at large-scale combats! Stop the presses!

D&D has always sucked at larger combats, so adding a rule that ends up discouraging DMs from running a large number of enemies seems fine to me.

Depending on your definition of large this sarcasm may be misplaced. The fact is D&D is actually pretty good at running relatively large combats, not mass combat mind you but large combats. My games very regularly feature combats with a few dozen opponents opposing the PC's and being able to roll 10 dice at a time for the Archer squad just looking for target numbers is actually a pretty fucking fast resolution mechanic. D&D has been able to handle sizeable combats for a couple of editions now, especially in comparison to the other major competitors of Shadowrun and WoD which DO use multiple dice resolutions which is what makes the difference.

hogarth wrote:It should be noted that your calculation of the mean is assuming that target numbers are uniformly distributed between 1 and 20 (e.g., that rolls that require a 1 or higher to succeed are just as frequent as rolls that require a 10 or higher to succeed and just as frequent as rolls that require a 20 or higher to succeed), which is a big assumption if you're talking about D&D.

Doesn't D&D generally assume target numbers closer to the middle, where the +3 generalization is most accurate? It's already been stated that +3 is an approximation. Should we also get pedantic on how a flat +3 bonus doesn't the mean the same when your TN is near the edges?

Yes, adding a reroll adds a bit of a curve to the RNG, which I was under the impression had already been stated so felt no need to be repetitious.

virgil wrote:Doesn't D&D generally assume target numbers closer to the middle, where the +3 generalization is most accurate?

On the contrary, +4 or +5 is more accurate in the middle, as I suggested above.

I'm not saying that it's a bad assumption to say that the target numbers are uniformly distributed (it certainly simplifies the calculation, which is nice), but it's worth pointing out.

virgil wrote:

hogarth wrote:It should be noted that your calculation of the mean is assuming that target numbers are uniformly distributed between 1 and 20 (e.g., that rolls that require a 1 or higher to succeed are just as frequent as rolls that require a 10 or higher to succeed and just as frequent as rolls that require a 20 or higher to succeed), which is a big assumption if you're talking about D&D.

Doesn't D&D generally assume target numbers closer to the middle, where the +3 generalization is most accurate?

Well, Next actually assumes player targets will be almost entirely within 12-17, at least in terms of AC, and 8-14 in terms of DC (saves against monster abilities). Against players, targets are assumed to be AC 12 to 18, but are more reasonably 16-22, because players will try to max this (trending toward 19-24 among players that really know what they're doing). And player spells will be more in the range of DC 13-20, but unlike everything else, that one is heavily level dependent.

What if Advantage/Disadvantage was +/- 1d6 to your d20 roll.

Is that better in any way to the rerolls that it currently is?

OgreBattle wrote:What if Advantage/Disadvantage was +/- 1d6 to your d20 roll.

Is that better in any way to the rerolls that it currently is?

Breaks the RNG more than a +3. Defeats the purpose.

OgreBattle wrote:What if Advantage/Disadvantage was +/- 1d6 to your d20 roll.

Is that better in any way to the rerolls that it currently is?

Worse, really. (3.5 rather than 3) And given the reaction to the old Next skill iterations, people really don't like it.

OgreBattle wrote:What if Advantage/Disadvantage was +/- 1d6 to your d20 roll.

Is that better in any way to the rerolls that it currently is?

Changes the variance and the expected value. Which is not a good thing for otherwise linear RNGs because that makes it really hard to calculate probabilities. Unless you're running dicepools (where doing this is mathematically tractable by eyeballing it) you generally only want to change the EV or rarely the variance.