Inferno target cap (nerf) inconsistent with other target capped abilities

Classes Warlock
1

The shard generation on the Inferno talent was recently capped to 5 targets (reduced generation past 5 targets). The problem with this nerf is that on 6 targets, all 6 targets have the reduced shard generation (rather than receiving full generation on 5 targets, and reduced generation on 1 target).

This means you receive significantly less shards on 6 targets than on 4 or 5 targets. This results in the gameplay that is less smooth as you add more targets (and reach the target cap).

Every other target-capped ability in the game does not work this way. Other abilities function in a way where you receive full value (damage) for the first X targets and deal reduced damage beyond that. This makes sense since you shouldn’t get weaker for adding more targets, which is currently the case for inferno.

2 Likes
2

Edit: Once I broke the math out more in the post below, the nerf makes more sense. I’ll have to experiment more to feel the extent of the nerf.

3

Thats the intention behind the nerf. Since we already generate shard to spam ROF at 5, keeping the “5” regen at 6 or more targets wouldn’t nerf the spec at all.

4

I imagine it was either this type of change, or they went full square root mode after a certain number of targets.

I haven’t tested or grabbed exact figures but let’s neckbeard some math here with 9% as a plug number.

You have 8 targets. Inferno generates full at 5 from 5, but reduced from all beyond 5 targets. Since there are 8 targets, you know none of these targets will give full % chance. If Inferno base is 20% and the nerfed one is say… 9%, you’re looking at 8*.09 = 0.72 (72%) for a fragment (rather than pre-nerf, which would have been 8*.2 = 1.6 (160%)). If that target count was increased to 20, you’d be at a very linear 20*.09 = 1.8 (180%)).
To recap:
8 * .2 = 1.6 [pre-nerf]
8 * .09 = 0.72 [post-nerf]

20 * .2 = 4.0 [pre-nerf]
20 * .09 = 1.8 [post-nerf]

If they went square root mode, you’d see something like this:
Assuming still 8 targets.
5 * .2 = A, 1.0 (100%) --this is still completely uncapped, sum of 5 targets.
(1 / sqrt(2)) * .2 = B, (0.1414213562) or ~(14.14%)
(1 / sqrt(3)) * .2 = C, (0.1154700538) or ~(11.55%)
(1 / sqrt(4)) * .2 = D, (0.1) or(10%).
A + B + C + D ~= 1.35 (135%)

But then the 20th target you’re hitting is
(1 / sqrt(20)) * .2 = E, (0.0447213595) or ~(4.5%).
The 40th target, if you’re MDI-style pulling
(1 / sqrt(40)) * .2 = F, (0.0316227766) or ~(3.2%).

TL;DR - if they wanted to target MDI style pulls harder, they would have put scale root resource generation scaling in place, but they opted for broader nerfs with a larger impact on the type of content you’ll more likely see.

If anything, this will definitely hamper our output because Inferno was making it to where you didn’t even have to bother with Immolate on pulls with 7+ mobs, BUT once you put a few Immolates on a pack and then start your ramp, it should be fine, esp with PI {pls delete PI blizz}.

1 Like
5

I could be wrong or just missing something as I’m no math wiz but I don’t think I can just sum the chances of it happening like that. There is always a chance you get no shards at all even with 100 targets at 20% each.

Don’t you have to take the odds of not getting them (.8) and raise it to the power = target number and subtracts that from 1 to get the odds of getting a shard per tick?

Unless I’m miss-understanding, your math gives the avg number of shards per tick for x targets.

6

You are correct. It would be (1 - 0.8)^(hits*targets). I could break down the probability further, but it’s much easier to follow conceptually by just using the base percentages. I don’t believe it’s the case, but there’s also the possibility that blizz uses pseudo-rng for the shard fragment calculation too.

I don’t know exactly what rng/pseudo-rng calculation is used, so it’s easier to just ‘flatten’ the figure into the average, so to speak. At least then it’s a number that users are seeing on the tooltip for a starting point.

If we say, 100 targets with 10 hits a piece in a true RNG situation, it’d be this:
A (chances of a fragment) = 0.2
B (no fragment) = 0.8
With 1000 totalhits, the probability of getting no fragments at all is… ~1.2302319221612 E-97 :man_teacher: :rofl:

1 Like
7

Blizzard has communicated precious little on the diminishing returns of Inferno, but I’m willing to speculate that a form of square root diminishing returns is in place.

Rather than count and sum additional chances for each successive target past 5, let’s take note of the fact that Inferno’s base 20% chance with 5 targets would average a return of 1 shard fragment every second. This shard fragment rate of return (SRoR) is the focus of their nerf.

For the sake of argument, let’s assume the nerf uniformly reduces the chance of individual enemy targets to trigger Inferno.

 If (number of targets) > 5
  SRoR = sqrt(0.2 * (number of targets)) 

  Inferno proc chance = SRoR / (number of targets)

Through this (theoretical) method, we could observe the following:

6 targets = SRoR 1.095445115, proc chance = 0.1825741858 or ~18.257%
7 targets = SRoR 1.183215956, proc chance = 0.1690308509 or ~16.903%
8 targets = SRoR 1.264911064, proc chance = 0.1581138830 or ~15.811%
.
.
20 targets = SRoR 2, proc chance 0.1 or 10%
40 targets = SRoR 2.82842712, proc chance = 0.0707106781 or ~7.071%

Applying square roots this way would ensure the average SRoR is never going to drop below 1, which would be an actual nerf. However, the rate of return will diminish with further targets.

As a general reminder, to anyone reading this post - this is speculation, not a statement of fact. I have no hard data to support my claims, instead offering this post as insight to how Blizzard might have implemented the Inferno nerf with minimal effort.

8

All I know is shard gen from RoF w/ Inferno feels terrible now. I even put immolate on every enemy and it still felt so slow. Considering going w/ Demo and the Implosion legendary now.

9

I will be honest here, with the results from MDI i would not be surprised if there comes another round of nerfs.