Dear Pro,
I wanted to jump in and give my two cents. There are a bunch of interlocking problems here, and “solving” some of them just creates new issues.
#1: Consistency of particular builds.
We all know that some builds are more consistent than others. A player of one build (S6 Impale, for instance), given a constant paragon and set of gear, might be able to clear GR X at their peak, and consistently clear GR X-5, seldom or never failing. Meanwhile, a player of another build (MOTE(6) Leapquake, or Wastes(6) WW, for instance) may also be able to clear GR X at their peak, but regularly fail GR X-8, X-10, or even X-12.
This is due to the ability of some builds (i.e. Impale) to deal significant damage to single targets, as opposed to requiring density (i.e. WW).
A build that relies on density runs into, and usually fails, quite a few “checks” in every GR they open. Say you’re playing WW. You open the rift, and see caves. You just failed. You can play the rift if you want, but your chance of success is 0%. So you remake the game, and try again… barracks. Failed again. The next one you open is a Festering Woods, but it’s full of Fallen Firemages. Another failure. Or, you could find a Festering with Transformers- a great level!- only to lack a conduit and thus be unable to kill many (or any) elites. Or, you could find the conduit, clear the level, get a great level two, only to spawn the RG and find that it’s one of the many you cannot kill in a reasonable amount of time without a power pylon.
Given the same rifts, the Impale DH vaults his way through, murders every elite he finds, kills a big pile of trash with a conduit, and then executes the RG in record time, with or without a power pylon. For this build, the “checks” are few. Density doesn’t matter too much. Pylons are nice, but not absolutely essential. And basically every RG is killable in good time.
This is why the DH GR clear average is so high: because of this one build. On the NA leaderboard, 91 of the top 100 clears, and 96 of the bottom 100 clears, are with Impale. The top 4 clears, however, are NOT with Impale (Top clear is 134, using N6 Rapid Fire, top for S6 Impale is 129). If you were to remove S6 from the game, the DH average would drop downward sharply, even though these top clears would remain the same. This is because N6 Rapid Fire, given optimal rift conditions, can take advantage of those conditions in a way that S6 Impale cannot.
Put another way, for these “density-seeking” builds, a bad rift produces an awful result, a good rift produces a mediocre result, and a great rift produces an amazing result.
For “single-target” builds like Impale, on the other hand, a bad rift produces a decent result, a good rift produces a good result, and a great rift also produces a good result.
When you combine these characteristics into one build, you get VyrTodo Wiz, basically the apex predator of the GR ecosystem. It’s tough as hell, annihilates density, kills single targets fast (especially when you start counting up Stricken stacks), and teleports quickly around the map.
In short, builds have a “spread”, between the highest GR you can consistently do (clear perhaps 75-100% of the time), and the highest you can do, period (clear 1 time). For “single-target” builds, this spread is low. For “density-seeking” builds, this spread is high. For builds that can do both, the spread is moderate, and the overall clear potential is very, very high.
This brings us to…
#2: Intra-class balancing, OR: "How good should that build be, and why?"
There are a few different conceptual schemes you could use for choosing which builds, for a given class, are better / more powerful.
-
You could just do it randomly- this largely seems to be the case now.
-
Or, you could stipulate that builds that are generally harder to play should be more powerful / clear higher. This is the approach Free and I adopted in our proposal. Simply put, it attempts to reward skill.
-
Or, you could stipulate that builds that take more time (i.e. fishing) to reach their highest clear should be more powerful / clear higher. This approach attempts to reward determination.
-
Finally, you could attempt to equalize all builds, so that none is better than any other. You could do this equalization along the lines of either average clear or top clear, but not both. Equalizing by average clear would result in builds with a high “spread” having the highest top clear (this is essentially the same as 3, above) , while equalizing by top clear would result in builds with high “spread” having a lower average clear.
These are your four options, there really aren’t any others.
Some people might seek to correct the issue of “spread” by either increasing the ability of “single-target” builds to deal aoe damage, or of “density-seeking” builds to deal more damage to single targets. And this approach is legitimate, up to a point.
Carried to its extreme, though, it leads to VyrTodo Wiz: a build that does everything well. When this sort of build exists within a class, it eclipses everything else. I have seen a lot of complaints from Wizards on this subject: that playing any other Wiz build seems pointless when you could clear higher, more easily, playing VyrTodo. And I do not relish the thought of other classes adopting this approach, because that would destroy the aesthetics and the feeling of differences in gameplay between one set, and one class, and another. So, I think (I hope) that to some degree, “spread” is here to stay.
For intra-class balance, I strongly prefer approaches #'s 2 + 3. Let the best builds (by a couple GRs, not a ton) be either the ones that are hardest to play, or the ones that take the most time. And try not to have any builds (like VyrTodo) that are the best at everything!
I would also say that “solo balancing, between classes” should not directly concern us (or it should be a very low priority). When you’re playing solo Barb, you’re competing with other Barbs, not with Wizards. The same is true for every other class.
In groups, however…
#3: Inter-class balance, OR: "Whose job is it, anyway?"
We need to talk about group play, and in group play, you can’t really divorce the idea of how good a build is from its particular job within the team.
When playing solo, you’ve got to do everything yourself: your toughness, damage to groups, damage to single targets, ability to group enemies, movement speed, everything-- is all dependent on you, your gear, and the skills you choose.
In a group, though, you’ve just got one main job, whether that’s providing survivability, grouping enemies, killing trash, or killing the RG. And while the ability of builds to thrive in solo play is somewhat reflected in groups, this is not a 1:1 comparison. Consider, for instance, the overwhelming dominance of VyrTodo in solo play, vs the equal dominance of Bazooka Wiz in groups.
Another example: in our proposal, Free and I imagined an upgraded Frenzy Barb (thorns or no) as a RGK. While this build might be inferior as a solo build to S6 Impale, at a certain point of damage and attack speed, it could stack Stricken fast enough to exceed Impale as a RGK. If you could somehow get to a 1 frame attack (60 attacks per second), you might surpass even Thorns Necro as a RGK, while still being considerably worse as a solo build.
Or, imagine if the Necromancer item Bloodtide Blade increased Death Nova damage by 4000%, rather than 400%, per enemy within 25 yards. A necro wielding such an item might well be a better trash clearer in groups than even Bazooka Wiz, but would probably have considerable problems clearing a solo GR as high as Bazooka Wiz can.
This is, of course, theoretical, but I hope the point is taken: in group play, specific builds have specific jobs, and being good at a meta job does not necessarily make a build very good at solo clears. Conversely, a build that is not very good at solo could still be excellent in groups (Free and I got an earful about this from Archael when we were working on the proposal, and I’ve taken his message to heart).
Adding buffs without acknowledging this fact risks massively deforming the overall shape of the game in unintended and unfortunate ways.
Conclusion
I have gone on far longer than I originally intended to, and have perhaps not expressed myself as clearly as I would have liked. For that I apologize. Anyway, here’s the bottom line:
I don’t think it’s a good idea to look solely at the overall averages, and work from there. That’s like looking at the finances of ten people, and thinking “well, on average, they’re all doing pretty well for themselves!”, without noticing that one is a billionaire and the rest are homeless.
Ultimately, I think balance should be based around:
-
Relative intra-class parity, either based on the difficulty of playing the build or the amount of time you have to spend to get a high clear.
-
Relative inter-class parity within group play, based on performance within meta roles (i.e. not based on an extrapolation of solo clear data).
We compete mainly against others in our class, and against other classes for spots in groups, and these two different competitions do not track evenly with one another. This means that there is essentially no average buff that can be applied across entire classes.
You’ve got to look at each build, assess how it plays in solo or in a group, and then make a decision about how that build should perform based on its difficulty of execution, its “spread”, the degree of fishing it requires, and its particular capabilities in group play.
Finally, I would say that a lack of absolute parity is OK. Some other class doesn’t need to displace, or even entirely catch, Wiz as a trash killer, or Necro as a RGK. Getting fairly close would be close enough. The goal is to have options, for most or all classes, that don’t feel massively overmatched.
Your data, Pro, is excellent and interesting, and I’m certainly NOT saying that we shouldn’t consider the averages, just that we shouldn’t consider them to the exclusion of all else.
Your Friend,
Rage