Fixing Rogues, Part 3: Cycle Complexity

In parts 1 and 2 of this series, I’ve provided fairly concrete suggestions for changes that should be made - specific changes to specific abilities that address specific problems that the class is having. And I have a fairly good-sized list of such proposals remaining; I could go several more weeks posting that sort of thing, and at the end of it the class would probably be better off than it is now. But having thought about it for a few weeks, I don’t think that’s the right thing to do. There are some major underlying issues that I haven’t fully solved, and its important to get the big picture right before dwelling too much on the details. So today I’d like to explain these problems and provide some rough ideas for addressing them. Not all of these ideas are necessarily good; but I figure the easiest way to figure out which are and which aren’t is to throw them out there, and see what people have to say.

The most fundamental goal of a rebalance such as this is to wind up with specs that are different and fun to play - once you have that right, everything else is at some level just tweaking the numbers to make things work. If the overall concept is sound, you can just pile damage multipliers on to get the balance right, as has been done with Combat (Vitality, Bandit’s Guile, etc.). Ideally the underlying concept should be strong enough that you don’t need too many of these - I think the fact that Combat gets about a third of its damage from such boosts is a sign that the underlying cycle is poorly designed - but so long as your underlying cycle concept is fun and interesting and at least in the right ballpark of power, getting the details of balance correct is easy by comparison.

“Fun” is, of course, relative; some people find Subtlety annoyingly elaborate and would rather play Assassination; others find Assassination boring and would rather play Sub. Clearly, we’re never going to be able to build specs that are all things to all people; but that doesn’t mean we shouldn’t try. I think a major part of this is getting the cycle right - complex enough to allow good players to distinguish themselves, without being unmanageably complicated or containing a lot of pointless busywork. Historically, rogue cycles have been sort of bad at this - it wasn’t until Wrath that you needed more than Sinister Strike and Slice and Dice to do 90%+ of the damage potential of the class. In more recent expansions they’ve tried to add some complexity to the cycles, but we’re still among the easier classes to play, and our cycles still aren’t really that interesting. So what went wrong?

The key, I think, is that rogue cycles are fundamentally repetitive in ways that most class’s aren’t. If you look at most other classes, when listing what moves to do when, they more often speak in terms of “priority lists” rather than “cycles”; at any given time you have some set of moves available to you, and you pick the best one based on what’s off cooldown, what procs you have active, and so forth. Because such rotations are based on the interplay of cooldowns of various lengths - coupled, perhaps, with random procs - the exact sequence of moves performed does not repeat on any meaningful timescale. Rogues, on the other hand, tend to perform the exact same sequence of moves with only minor variations many times within a fight. Ignoring finishers with a duration (SnD, Rupture, etc.) for a moment, in general there is a best finisher to do, and an ideal number of combo points to do it at, and an ideal way to get that number of combo points. Assassination rogues Mutilate to 4+, and then Envenom. Combat rogues SS to 4+, RvS if necessary to get 5, and Eviscerate. Some degree of complexity can be added by adding additional steps to this rotation, but fundamentally: these steps get repeated hundreds of times per raid, and there are only so many different ways to count to 5. Even if you had a different strike to use at each number of combo points and could remove combo points as well as adding them, I suspect you’d quickly wind up with a cycle where the hardest part of doing it is remembering what you’re supposed to do - and to me, that doesn’t sound like much fun.

To fix this, we need mechanics that change what is optimal over time, such that every time I’m at 3 CP I don’t necessarily do the same thing. The simplest way to do this is to introduce abilities with durations, so sometimes I build combo points for an Eviscerate in whatever the optimal fashion is, but sometimes I need to build combo points for a Rupture or a Slice and Dice instead. What move I do next is no longer purely a function of how many combo points I have; its also a function of how much time is left on my various timers, and as my energy and combo point generation is somewhat variable, the circumstances will be a little different every time. However, this is not without its own set of problems, foremost of which is that sort of by definition, such abilities provide their benefit over the next 15 or 20 seconds, thus exacerbating our existing burst problem. And, of course, if all specs rely simply on juggling DoTs for cycle complexity, the specs will not be well differentiated. Hence, while we can use timer juggling for some trees, we don’t want to do it for all three; and I think the natural choice for finding “something else” is Combat, simply because it has a harder time solving the burst issue than the other specs.

Rogues, in general, cannot solve the question of burst like other classes do. We have no notion of long-term efficiency, so we can’t trade efficiency for short-term burst like an Arcane Mage; and the need to build up for finishers means we can’t reorder moves to line up our hardest-hitting moves together like a Frost DK. Hence, the easiest way to get burst is by adjusting the spacing between our moves - we don’t have the energy to do moves in quick succession all the time, but we can save up a bar of energy and do 3 or 4 in quick succession, which for something like shields on Conclave, is all you need. However, this burst will only be meaningful if our energy regen and move cost is such that we normally have significant gaps between our moves, such that removing them gives a significant short term benefit. To make this work, you need relatively low energy regen and a fairly expensive combo point generator. However, low energy regen and an expensive combo point generator means low combo point generation, which is not ideal. Hence, this approach requires either a) a combo point generator that gives more than one point per cast or b) some external means of bonus combo point production. Or, as these solutions are more commonly known, “Mutilate” and “Honor Among Thieves”. Hence Assassination and Subtlety are in a good position to get a working DoT/self-buff based cycle - the numbers certainly need some tweaking, but the underlying mechanics are mostly there. Combat, on the other hand, is not well set up for DoT juggling. SS is cheap, energy regen is high, SnD is extremely long, and Rupture is a marginal DPS gain at best; none of this works for an interesting DoT based cycle. And since we want at least one spec doing something different, Combat seems like a natural choice.

Now, Bandit’s Guile is clearly an attempt to do this - its a periodic buff that works off a different timer than the regular combo point cycle. However, the fact that its completely uncontrollable and affects all damage equally means it doesn’t change our cycle in any meaningful way; what’s optimal at Shallow Insight is (generally) still optimal at Deep Insight. Hence, while its not a bad idea in theory, the implementation is a complete failure - we need to find something different, and better.

With these problems in mind, lets take a look at some potential solutions.

I. Finisher Size needs to matter more.
This one is almost a no-brainer - the only reason its in this article of speculative solutions is because while the *what* is easy, the *how* remains elusive. Right now, the benefit provided from a finisher scales linearly - if that - with the amount of damage it does. If you look at the actual damage done by the finisher, a 5pt finisher almost always comes out behind doing a 3pt finisher and a 2pt finisher - the only advantage it has is that it costs less energy. And as Blizzard’s stated goal is that we build up potential to unleash powerful finishers, not build up potential to do damage in a slightly more energy efficient way, it seems like this should change - particularly since it would also make timer juggling more interesting. Right now we can get away with refreshing with whatever combo points we have on hand - the loss of efficiency is small enough that is not worth worrying about. The solution to this is to make larger finishers do more than proportional damage relative to smaller ones. While Envenom right now does n/2n/3n/4n/5n damage based on the number of combo points you have, it should instead do (say) n/2n/3n/5n/8n damage, so that it’s more important to build up to the higher levels. And while such adjustments are fairly straightforward for Eviscerate and Envenom, its a little less clear for some of the others; should SnD merely get more than proportionately longer, or should it also give more attack speed at higher levels? Is it sufficient for Rupture to do extra damage per tick, or should its duration also increase more than proportionately?

II. Garrote can be used out of stealth

• Glyph of Garrote changed to: Prime Glyph. Garrote can be used outside of stealth, but no longer silences the target.
• Venomous Wounds proc reduced to 6 energy.
• New 1pt talent which allows Garrote to be used even when not behind the target.

Something needs to be done with Garrote for Assassination. As discussed in my last post, Vanish should not be a DPS cooldown, and while removing Overkill solves most of the problem, as things stand one would still Vanish for Garrotes. One solution would be to simply remove Garrote from Venomous Wounds; but I think a more interesting solution is to allow it to be used out of stealth, such that it becomes a regular part of - and source of complexity in - our rotation. To prevent giving rogues a second, spammable interrupt, we then need to remove the silence from it, which more or less requires the involvement of a Glyph. We also need to cut the energy given by Venomous Wounds to keep overall regen under control, and providing a way of reducing the positional dependence seems like a good idea, and makes a nice utility talent. All in all, there’s a lot I like about this idea.

However, its not without its problems, either. In addition to the burst issues, is also further reduces the value of Expertise - coupled with other forthcoming changes, Expertise would become our dump stat, which isn’t necessarily ideal. It also seems odd to gain an essential part of your cycle through a Glyph, and the Glyph winds up being significantly more powerful than all other options - its fine if its clearly the best, but being better than the next 3 options put together seems excessive. And finally, its not at all clear to me that this alone injects enough life into the cycle to keep it interesting; would the existing cycle with Garrotes woven in every 18 seconds be sufficiently more interesting than the current setup, or do we need to find something better?

III. Subtlety drops Slice and Dice.
I’ve always felt it would be interesting to try to mix up which finishers each spec uses a little more, and Subtlety strikes me as a very natural candidate for breaking the SnD monopoly - removing it from the Subtlety mastery, coupled with a much-needed global nerf of SnD probably gets you most of the way there, given the large amount of damage dealt by Sub’s other finishers. The problem is, this turns the Sub finisher rotation into an exact clone of what Assassination currently does - you have one duration-based finisher you hard cast every so often to get bonus energy regen, and a second one that you cast once at the start of the fight and refresh thereafter with your direct damage finisher. This is kind of dull. So to make this work, we should really nerf the rupture refresh mechanism. If, say, each Eviscerate extended the duration of your current Rupture by 2 extra ticks instead of giving a full refresh, you’d still have an interesting and beneficial talent, but you’d need to pay more attention to Rupture - which strikes me as a good thing. I think this is an idea with some interesting aspects to it, but would it actually improve the spec? You tell me.

IV. Rethink HAT
I admit I’m not a huge fan of HAT in general - call me old-fashioned, but I miss the good old days where combo points only came from hitting buttons. But, as noted above, it is an elegant solution to keep combo point generation high, as well as allowing raid DPS to be competitive without impact small-group PvP, where the spec is already strong - hence, its probably worth keeping around. The only problem is: the combo point generation is too regular. In any raid scenario, you gain a combo point pretty much *exactly* every 2 seconds, and that regularity and reliability means the cycle isn’t as interesting as it could be. If your bonus combo points came a bit more randomly, with droughts and floods that you had to react to, it seems to me that the spec would wind up more interesting as a result, even if the average combo points gained in practice was about the same. So how do we do this, while keeping the nice scaling properties of a larger benefit in a larger group, without reverting to the broken HAT of days of yore?

...I haven’t the faintest idea. But I still think its worth thinking about. If you have any suggestions, I’d love to hear them.

V. Rework Bandit’s Guile to be more like Eclipse
As noted above, one of the major problems with Bandit’s Guile is that it boosts all your damage equally. What if it didn’t? What if, say, it only boosted your magic damage, but not your physical damage? And, further, what if we also had a move - lets call it Shadow Strike - that is similar to Sinister Strike, except it does Shadow damage instead of physical? Assuming proper balance, we could wind up with a situation where at low stacks of Bandit’s Guile, you use Sinister Strike, and at high stacks, you use Shadow Strike. It could even be balanced to make you switch between Eviscerate and Envenom as well. This would clearly be a bit more varied than what we currently do, and it’d definitely be a step in the right direction. But we can do better still by addressing the other major problem with Bandit’s Guile - its uncontrollability. This is where the Eclipse comparison kicks in.

Eclipse, to give an oversimplified summary for those of you that aren’t familiar with Balance druid mechanics, is a mechanism where performing Nature spells makes your Arcane spells better, and performing Arcane spells makes your Nature spells better. While its a bit more complicated than that in practice, I think we can adapt this general principal to spice things up for rogues. Instead of Insight being purely something that stacks up and drops, lets make it stacks of a (de)buff that gives, say, +3% physical damage, -3% magic damage (or the reverse). As we do more Shadow Strikes, we gain stacks that increase our physical damage at the expense of our magic damage; as we do more Sinister Strikes, we lose those stacks and gain ones that increase our magic damage at the expense of our physical damage. And lets further make Shadow Strike always give 1 combo point, as opposed to Sinister Strike which has a chance to give 2.

Now, it seems to me that it should be possible to balance this in such a way that its “normally” optimal to keep your stack somewhat balanced - say, to keep your +physical buff between +3 and +12%. You’d use about the same number of physical and magical attacks to keep your buff in this range, while also exploiting their different combo point returns to minimize the number of combo points you waste (sort of like you use RvS at current). You wouldn’t purely alternate the two moves - you’d need to response to Glyph of Sinister Strike procs and do a few more or a few less Shadow Strikes to make things work out. Furthermore, you’d gain the ability to sacrifice some damage in the short term in exchange for long term benefit by stacking your buff all the way up to its maximum (whatever that happens to be) in order to get a significant string of highly buffed Sinister Strikes in a row - thus giving you significantly more control over when you do more or less damage than is currently possible - that is, burst.

Now, I’m sure there are many problems with this idea as well, as well as a lot of picky details necessary to get the balance right. I make no claims that this is the ideal solution - or even a good one. But I do think its an interesting solution, and thus worth consideration - it is going to take something of this magnitude to fully fix Combat’s broken mechanics. So feel free to tell me what you like, what you don’t like, or what you think should be done instead; as until we get the big picture sorted out, the details don’t matter.