Monday, May 31, 2010

Power Word: Shield (Rank 12)

In Burning Crusade the developers did a lot of tinkering with downranking. When the expansion first came out, downranked spells were super fantastic. While spell power accounts for a lower percentage of healing than it does for damage, lowering the base effect to lower the mana cost was a trade people were all too happy to make. In fact, it hardly made any sense to cast your most powerful rank of a healing spell in any situation, since they generally cost quite a lot more and only healed for a comparatively small amount more. They put in some arcane calculations to make downranking worse by lowering the spell power coefficients of spells that were learned at levels lower than a character's current level. These occasionally even lowered the spell power of your highest rank of a spell, since sometimes the highest rank was learned more than 5 levels before you hit max level. In Wrath they decided to kill downranking once and for all by making lower ranks of spells cost the same amount as higher ranks.

Despite this, I've been casting a lot of Power Word: Shield (Rank 12) recently. With the ICC buff and a fully powered Demonic Pact I could be shielding around eleven thousand damage. Or I could cast a shield two ranks lower and shield around seven and a half thousand damage. Same cost, just a worse effect; that's what I'm doing.

Discipline priests get a lot of mana return from Rapture. Rapture gives back 2.5% of the caster's maximum mana every time one a shield breaks. It does not give back any mana if the shield expires without breaking, even if the shield only has 1 point of absorb left. This effect can only occur every 12 seconds, but if multiple shields break at precisely the same time, the 12 second cooldown is unable to trigger in time, allowing all of them to return mana. Normally you wouldn't even consider this, as 30 seconds is a long time for a shield to absorb all the damage its going to absorb. But there are a number of factors at play that make downranking the right move on Lich King.
  1. Infest does just a little less damage than is required to break a full powered shield
  2. Because of the mechanics of infest it is not acceptable to let one shield cover two consecutive casts; each infest must hit a fresh shield
  3. The lich king has no other splash or AoE damage during phase one
  4. The mana return from Rapture is necessary to keep up with the healing demands of the fight
Heroic 10-player Lich King does 11.7k to 13.2k damage with Infest. This is reduced by resistances, which average 20%, and reduced by Renewed Hope, making the average hit 9680. If I cast a fully powered Power Word: Shield then the shield would hold against this damage. The only way the infest breaks the shield is if it rolls near max damage (13,218) and the victim gets only a 10% resist (about a 25% chance). This will happen sometimes, but much less than most of the time.

Infest from heroic Lich King is a serious problem that the healers don't have a lot of leeway to deal with. Allowing the debuff to be applied to more than one or two non-tanks probably means it will be fatal, which means a wipe. If someone takes 9680 through a 7300 point shield then they end up taking 2380 damage. As this is less than 10% of their maximum health, they infest debuff will fall off immediately. If the same person gets an 11k shield then the first infest reduces that to a 1320 shield. That means the next infest will do over 8000 damage to them on average, and they will need heals to clear the debuff unless their total health is over 80k. Because your buffer to avoid the Infest debuff is your shield + 10% of your health, the two smaller shield approach generally allows the target to take more than 10.5k and still not get the debuff. The single large shield approach, however, doesn't get to use the 10% health buff from the first infest. In order to succeed at preventing the infest debuff from two consecutive infests it would need to be enormous, in excess of 16k shielded. This won't be reached even with a 30% ICC zone buff and level 277 gear.

Therefore a shield is needed for each non-tank for every infest. In addition to that, there is no real difference between someone coming out of an infest with 90% health and someone coming out with 100% health. Judgement of Light or other splashing heals will heal them to full before the next infest. Given that, the larger shield doesn't actually protect any better than the smaller one.

Finally, the mana cost of healing lich king is extreme. Infest happens every 23-25 seconds, usually closer to 23. With eight shields per infest, that means spending over 230 mana a second on shields alone. Since nearly every GCD is spent on one of Renew, Penance or Flash Heal, mana spent per second on this fight is probably around 500 for priests. Discounting Rapture, 2500 mp5 is not an attainable goal. If my shields never break to infest, that is what would be needed to keep up with the healing. With shields breaking on every infest, that's over 800 mana times 8 shields every 23 seconds. By downranking, Rapture is providing about 1400 mp5. It is also providing 2% mana every 23 seconds to other casters in the raid (about 40% of a replenish) and rage, runic power and energy to the non-mana users in non-trivial amounts.

Rapture has has this problem for a long time, but for a long time there was no circumstance where it could rightly be called a problem. The developers did this because they were solving another problem: they didn't want discipline priests to simply cast shields for mana regen when they were running around. Rapture is only supposed to give you mana back if you are actually using the shield to prevent damage, not if you are casting it just for fun. I can understand that goal. They didn't think much of the problem that having shields that were too large could prevent you from getting your mana back. After all, what were the odds that discipline priests were going to start complaining that their shields were too large?

It took a very strange confluence of events to make discipline priests worry about their shields being too large. But it just happened to happen on the encounter that is the culmination of the entire expansion. It also happened to happen on an encounter where discipline priests are favoured (some would say required) for healing. While resto druids are complaining that they don't have the tools to heal infest, discipline priests are busy intentionally reducing their best spell by a third of its power to properly deal with infest.

Is all of this a mere quibble? What difference does it really make if I downrank on one fight because of a bizarre mechanic?

The problem it causes is that it makes how powerful I am seem irrelevant. If I had more spell power I would just have to drop another rank of Power Word: Shield, and having the wrong amount of spell power might be a disaster. More crit wouldn't be benefiting that task at all. I spend nearly the entire fight under the effects of Borrowed Time so my haste hits a soft cap around 154 rating and does very little after that point. With the massive mana regen from Rapture and Infest I have no use for additional Int.

I've also considered redoing my spec to get rid of Improved Power Word: Shield. But what would I take in it's place? The lich king has no effects that are affected by Unbreakable Will. Reducing my threat with Silent Resolve would similarly have no effect. Should I put two points in absolution to save myself 250-300 mana over the course of the entire fight?

There may be a non-facetious argument in favour of switching my gems to stamina, a bizarre situation to say the least.

My gear doesn't feel like it's making me better, my talent points are wasted in talents that aren't helping just because the alternatives don't help either, and I am intentionally throwing away a third of the effect of my most powerful and often cast spell. Time to go take on the hardest boss in the world!

Wednesday, May 26, 2010

Spell Power Coefficients & The Real World

Disclaimer: I run strictly run 10-player raids. The point I am making here is a point about how spell power coefficients distort how the game is actually played and the decisions we make. I believe the particular example doesn't necessarily work in 25-player raids.

Recently I was reading a blog post at Penance Priest about how to heal as a discipline priest. The central theme of the post is very good advice, and advice that actually applies very well to most classes these days, not just discipline priests.

But I'm not really going to comment on that post, I'm only bringing it up because it reminded me of a recent change I've made in the way I play. The post casually mentions that Renew is terrible. And it's natural that it mentions this casually, as a sort of passing joke, because every discipline priest knows that Renew is a bad spell for discipline. We know this because we started this expansion with around 2k spell power. Renew heals for:

(1400 + (1.88 * spell power)) * 1.05 [Twin Disciplines] * 1.04 [Focused Power]

That's 5635 healing with 2000 spell power. You could get it to 6480 if you took Improved Renew, but Renew is bad, so you don't take Improved Renew.

At the same time, Flash Heal heals for for:

(2040 + (0.806 * spell power)) * 1.04 [Focused Power] * (1 + crit chance * .95)

Flash Heal healed (and shielded with Divine Aegis) for 5241. So if you are willing to wait 15 seconds for your healing you could get about 7.5% more. This is not much of a trade-off. Even if you ignore crits and spend the talent points on Improved Renew you would need three Renew ticks to equal a Flash Heal.

But we don't have 2000 spell power anymore. In fact, we can have more upwards of 4500 with a demonology warlock in the raid. With 4500 spell power Renew heals for 10800 while Flash Heal heals for 8130. Now if we ignore crits and spend the talent points on renew we need 2.4 ticks to equal a Flash Heal. Renew improved about 25% relative to Flash Heal over the course of the expansion.



Even compared to Power Word: Shield, Renew is still looking like something you might want to cast sometimes (given that Weakened Soul is a thing). I've mentioned before that I realized at some point that I didn't want to be using shields to raid heal on Blood Queen because I want to save them for pacts. What do I do with my GCDs when I am not shielding? I was casting Flash Heal at first, but now I cast Renew. What do I do when one healer is ice tombed and the other one has unstable magic on Sindragosa? I cast shield on anyone at low health and pre-emptively throw renews around. What do I do when it's almost Bonestorm time on Marrowgar? I put Renews on people and keep shields for when I see who actually needs them.

In other words, I behave the way that I think the developers imagined I would behave when they were designing things. Renew is a preemptive heal against damage I know is coming, Penance is to bring someone's health up, Shield is for emergencies and to deal with attacks that do enough damage to actually threaten to kill people in a couple of seconds. The nature of the spells, ignoring the particulars, suggests this is how you would want to play.

It is the particulars that made Renew useless. Power Word: Shield is still a better spell, and Flash Heal still has lots of use. But with sufficient spell power, Renew has lots of use as well. Discipline priests rarely use Renew because we all took it off our bars so long ago. But as spell power has over doubled in the expansion, the value of Renew has radically changed compared to the alternatives. As I've said before, the coefficients of spells should bear a resemblance to the power - or intended power - of the spells, not to the duration or cast time. The result of the current system was the extreme distortion of the power of Renew and a "generation" of discipline priests who don't look twice at a meaningful part of their toolbox.

Friday, May 21, 2010

13%

Some recent cataclysm previews have shown that the current 13% spell damage buff is going to be dropped to 8%. While this is a step in the right direction, they need to take another step before they are done.

As I pointed out, the 13% spell power buff gives approximately 8.5% more damage to a typical raid of mixed dpsers while the next best buff is around 3.2% extra damage to the same raid. But decreasing 13% to 8%, other things being equal, it will now give a 5.2% damage bonus to a typical raid, so about 63% more than the next best buff.

If they dropped it to 4% instead then it would be a 2.6% buff to raid damage, making it well above average in terms of raid buffs. At 5% is would still be the most powerful raid buff, but by a very small factor instead of by a very large one. I hope that cataclysm reduces this buff to 4% or 5% and gets it under control so that a raid without this buff can function, much as a raid without any other particular buff can.

Thursday, May 20, 2010

Crit Bad. Damage Good.

My post about rating conversions contains the seed of an argument that I'd like to spell out in more detail. Dps classes in PvE are balanced based on how much dps they do, not based on how often they crit or how much they crit for. If one class crits for double damage and another crits for triple, that doesn't really make a difference to class balance. The class that crits for more will either crit less often or have lower non-crit damage to make up for the difference so everyone is equal1.

So lets say that in cataclysm dungeon blues the developers are aiming for about 10k dps. Let's also say they are aiming with current talents and rating conversions for us to crit about 20% of the time. That means that if you crit for double then without crits you would be doing 8333 dps. At 8333 dps, when you crit 20% of the time, and crits double you damage, you end up doing 80% regular hits (.8 * 8333) and 20% critical hits (.2 * 16666) and that adds up to 10k dps (ignoring the rounding I've done). If you crit for triple damage than without crits you do only 7142 damage. Again, you have non-crits 80% of the time for 7142 damage and you crit 20% of the time for 21428, averaging to 10k.

If you gain 1% more damage from a talent or a buff, then we know how much damage that is. Regardless of what class or spec you are, in our simply and perfectly balanced world of dps, 1% of 10k is 100. A 1% boost in damage means you do 1% more whether you crit or not and it always comes out to 100 more dps.

What if you crit 1% more often? Well, if you crit for double then your it means you do your base damage of 8333 79% of the time, while you do your critical damage of 16666 21% of the time. .79 * 8333 + .21 * 16666 = 10083 dps. So 1% crit means only 83 more dps. If you crit for triple, of course, crit is better, it gives you .79 * 7142 + .21 * 21428 - 10000 = 142 extra dps, making it better than dealing 1% more damage.

The break even point between these two values, where 1% crit is the same as 1% damage is critting for 2.25. At that point 1% crit is worth 100 dps just like 1% damage.

The math here might be slightly unintuitive. If you crit for times two damage then each 1% crit adds 1% of your non-crit damage to your final damage. If you crit for times three then each 1% crit adds 2% of your non-crit damage to your final damage. But in reality critting for times three doesn't make crit twice as good as critting for times two. This is because, in order to arrive at the place where you do the same overall dps, your non-crit gets smaller as your crit multiplier gets higher, given a fixed crit chance.

What happens to the same calculations when you have a 25%, 30% or 35% chance to crit instead of a 20% chance to crit?

In the case of critting for triple, the amount of extra dps you get from critting goes down from 143 to 133 to 125 to 117 as you increase your crit with the same dps number. At 50% crit, a value that is probably a little low for today's dpsers, critting for triple is just enough to make 1% crit as good as 1% damage. Any more than 50% as an expected crit value and you'll need to crit for more than triple to make a crit buff as good as a damage one.

But very few classes actually crit for anything like that much. The physical dps classes get pretty good crit multipliers because they start with times two and so their talents tend to put their crit over that. For casters, though, they start with critting for only 50% extra and their talents tend to take them up to double. The most extreme example of critting is the now defunct frostfire mage, who crit for 2.25 and then put a DoT for 40% of that on the target, making their total crit over triple. Fire mages crit for only 2.45. Hunters can also get pretty big crits with Aimed shot, critting for up to 2.4 with 30% of that becoming a bleed for a total of 3.12.

Almost every class, however, has a fraction of their damage that cannot crit. Almost every class has talents that boost the crit of some but not all of their damage. In the case of hunters Aimed Shot and some other shots might crit for 3.12 (3.21 with the metagem) for Marksmanship spec but their auto-shots and wild quiver shots crit for only 2.3, and those account for a significant portion of hunter damage. This lowers the effective bonus damage from each percentage point of crit.

Developers are concerned about giving out too many crit buffs because crit has interactions with other abilities. But in reality there are few classes that benefit more from a 1% crit increase than they do from a 1% damage increase, and as an expansion wears on the number of classes that benefit more from a 1% crit increase than a 1% damage increase will get smaller and smaller.

Of course this only matters for weighing buffs, talents and glyphs against one another, but it is something that I hope the developers can realize and keep in mind when they are balancing Cataclysm.



1. Everyone is not equal, but assuming they are supposed to be roughly equal, we can go forward.

Tuesday, May 18, 2010

Enchants

I've made posts about gems and glyphs, now it's time to move on to enchants.

Just like gems and glyphs, enchants are great things to get. My main is an enchanter, and while my other profession moves around a bit, I can't ever imagine dropping enchanting. But enchanting really isn't what it could be, and much like gems and glyphs, this has a lot to do with the selection of enchants that are available.

Looking at enchants for various slots, there are generally three or four different "top level" enchants for each slot, that is, enchants that give the maximum possible number of stat points. In gloves the expansion brought both Expertise (15 expertise rating) and Precision (20 hit rating). While given that choice (and only that choice) some characters should take Expertise, the Hit rating is the more powerful enchant.

In gloves you get to choose between AP, Spell power, Agility, Hit and Armsman.
In cloaks you get to choose between Agility, Haste and Armor
In chests you get all stats or Health
In boots you get Tuskar's Vitality unless your class gets to run fast, in which case you can choose between Agility and AP
In bracers you get to choose between Stamina, AP and Spell Power

The question is, are there any interesting choices to make here? In reality, the closest any class comes to an interesting choice is knowing whether AP or Agility is better for your the spec you are in. Probably the only place where enchants are really interesting is where one stat is so brutally unbalanced or desperately needed that it actually makes sense to take a hit to your stat points in order to convert them to that stat. For example, if one strength is worth three or more AP, as it was for Unholy DKs for at least part of the expansion, then you should get the Major Strength enchant from BC on your gloves instead of getting Crusher.

In reality aside from the agility/AP classes, the choice of enchant is a transparent non-choice for even a novice player with no theorycrafting experience at all. If an arcane mage has to choose between Haste and Crit then there is a right answer and a wrong answer for maximizing damage, and so you can say that the choice can't really be all that interesting. But a novice player has the option of customizing; diminishing returns on one or the other stat might affect their relative weight as your gear changes (for example, Crit might proc an ability like Flurry and be less useful per point when you have 40% than when you have 20% because of the chance in how it affects your Flurry uptime); and at least in order to figure out which stat you want you need to know something about your class mechanics. Choosing between Spell Power, AP and Stamina on your bracers isn't a real choice for anyone.

Enchants that do more than simply add a stat make enchanting more interesting in a way, but often make choices less interesting. Armsman is the tanking glove enchant. It's hard to imagine a tank using another enchant (aside from tinkering armor on with engineering), and no non-tank would use it under an circumstances. Tuskar's Vitality fills the moving faster gap for specs that can't talent it, and I'm not even interested in arguments that a small number of stat points can be better than 8% faster movement. Those arguments have played themselves out already. Except in cloaks (where people don't use the "special" enchants) there is only one special option per slot, and that special option is either useless or overpowered depending on your spec.

I don't think that increasing the selection of enchants is really going to do the trick. Each class has nearly 30 major glyphs to choose from and there are about 50 epic gem cuts, but all that variety doesn't increase the breadth of rational choices very much at all.

Changing enchants so that we choose between stats that we might actually want to choose between would help, though. If I could choose hit, haste or crit to wrists then my choice would depend on my spec, my current gear, and to some extent the fights that I was facing, instead of depending only on my spec.

Adding more enchants with "special" effects that are useful in a wider variety of circumstances might also improve things. Engineering shows that they have the capacity to do enchants that give items use abilities. More enchants like this, where you get to choose between an on-demand boost and a constant passive boost, would give more leeway to choose enchants based on a style of play and on fight mechanics rather than simply choosing the one that is best for you on a spreadsheet.

Of course there are some major obstacles to overcome. If spell power is as out-of-whack with crit as I suggested it would be in my recent post of rating conversions then in order to make a glove enchant tempting for a mage, they might have to make it give 140 or more crit rating, since otherwise the WotLK spell power enchant would beat it out. I don't expect crit rating to be that far back from spell power in reality, but I wouldn't be surprised if it were no better than a third of Intellect for most caster classes when Cataclysm arrives. Old enchants would be taken over new ones because stats are so desperately out of scale with one another.

My prediction on this is that enchants will basically be the same as they are now when cataclysm comes out. All the casters will enchant Intellect and lament it when they can't. All the tanks will enchant Stamina and have similar complaints about that option not being available on every slot. I don't think enchanting is quite as degenerate as glyphing, and neither are even close to gemming, so I expect it to get the least attention. And all of these things have to compete with their ultimate goal for customization: making talents into real choices. If they can get that right then they can probably rest on those laurels of character customization for the entirety of cataclysm and not worry so much about the homogeneity of enchants, gems and glyphs until next expansion.

Tuesday, May 11, 2010

Rating Conversions

Well, we've gotten our hands on the level 85 rating conversions. It looks like you are going to need 268 crit rating per point of critical strike percentage.

Of course the numbers on all of your gear are going to inflate quite a bit: you'll have lots more critical strike rating at level 85 than you do now. Still, requiring about four times as much rating per point of critical strike is going to mean less critical strike, because your stats are not going to be four times what they are now.

We have seen that the new enchants need a level 300 item to work. This is from the alpha and could be way off, but level 300 is a pretty good prediction. Between vanilla and BC first tier raids we went from level 86 to level 115. Between BC and WotLK we went from 164 to 200. Top tier WotLK raids drop 277s, but we know things got a little out of hand in terms of item levels, so between 300 and 315 probably what we are looking at for item levels for first tier cataclysm raiding.

In first tier WotLK raiding we had about 2000 spell power, so an arcane mage derived around 55% of their damage from spell power and the remainder from base spells. A well-geared (not top-of-the-world guild but doing okay) mage these days has around 3500 spell power and 1000 crit and haste ratings. If we go from 264 to 300 gear we can expect about a 40% increase in our stats.

Let's look some numbers. Please bear with me through these loose estimates, the accuracy of the estimates will end up making very little difference to the conclusion. We will guess a level 85 mage with first tier raiding gear will have around 4900 spell power (pre-buffs) and around 1400 crit and haste ratings. Let's give them a 5% base crit and estimate 15% crit from talents and buffs. That would give them a total crit of around 25%. We'll use the first-tier WotLK value to estimate how much of their damage comes from spell power and how much comes from the base spell. And lets, for the sake of argument, say the mage does about 15000 dps (based on the fact that mages can do almost this much right now and that damage between the end of BC and the beginning of WotLK remained pretty similar after as rating deflation cancelled out increased spell power).

With all this, 1 spell power would increase the mages' dps by around 1.7. Arcane mages crit for 1.75 times the base spell, so how much would 1 crit rating increase their damage. Well, one crit rating would increase their crit chance by just under 0.0039%. That means instead of doing 75% extra damage 25% of the time, they do 75% extra damage 25.0039% of the time. 75% of their damage is 11250, and they do it (1.250039/1.25) - 1 = 0.00312% of the time. That's 0.351 dps. That makes crit rating worth about 20% of a spell power.

Now spell power won't be on items, and we don't know exactly how you will get it, whether it will come 1:1 from int or at some other ratio, but it is pretty reasonable to expect that we won't have a great deal less spell power at 85 than we do at 80, so we are going to need to get it at roughly the same levels as we do today. At worst I can imagine a 1:1 ratio from int, meaning that spell power would be slightly more expensive item point wise than it is right now. Even if it is 1:1 (rather than it's current .85:1) that would mean having a stat that is at least 5 times as good as another for doing damage. Of course mages might have talents that increase the amount they crit for or proc off crits in other ways. But in order to make spell power and crit even out they would need to crit for the equivalent of five and a half times their normal damage.

Blizzard has recognized that it creates problems when one stat is so much more powerful than another, and I'll be very interested to see how they get around this.

The most your crit can get to is 100%, and they really don't want it to get anywhere near that high. You have a base amount and you have buffs and talents, so your crit probably ends up being 20%-ish before you get any crit rating at all. With over half your damage coming from spell power (and arcane mages actually have a pretty low ratio here, with 2000 spell power fireball is well over two-thirds spell power and Mind Flay is three-quarters spell power) you can expect your damage to increase by 30%-ish over the course of the expansion based on spell power increases (this would equate to a 60% increase in spell power). Suppose you start the expansion with 20% crit from base/talents/buffs and 5% from rating. In order to increase your damage by 30-40% from crit rating you would need to go from 25% crit to 62.5% if you crit for double. Since you started with 5% crit from rating, that means increasing your crit rating by 8 and a half times. Even if you crit for triple you would need to increase your crit rating by four and a half times to achieve that kind of damage increase.

How much would you need to crit for before a 60% increase in your initial critical strike rating would result in a 30% increase in damage? Fifty times! That's right, with a crit of 20%, 5% of which is from rating and 20% of which is from talents, assuming that a 1% increase in spell power is equivalent to a 0.5% increase in damage, it would require that you crit for 50 times normal damage for you to value a 1% increase in spell power the same as a 1% increase in crit rating.

What kind of solutions could there be? There are two breeds of solutions to this problem, I think. One is to take steps to equalize the stats, the other is to make it not matter as much that stats are so different in value.

In order to equalize the values of stats, they need to pound base crit value into the floor. Every point of base crit value has a debilitating effect on crit rating. There is the minor effect of reducing the actual effectiveness of each point of crit percentage you get (it means less as a percentage of your total damage to move from 150% to 151% than it does to move from 120% to 121%). But more importantly each point of base crit value eats into the cap they eventually want you to get to. If they envision you getting to 30% crit and you start with 20% then all the crit rating you ever get in the expansion can only give you 10% crit. If they envision you getting to 30% crit and you start with 10% then they can halve the crit rating conversion, doubling its power relative to spell power. Another thing they can do is make base spells account for much more of our damage so that spell power doesn't improve damage as much as it does now.

In order to make the extreme difference in stat values not matter so much, they can make it so we cannot gem or enchant our favourite stats. I've talked about why gems don't work in practice before. If one stat is even 50% better than the alternatives then the odds you gem anything but that stat get pretty low. If one stat is five times better than the alternative then obviously you gem nothing but that. If they don't let us gem or enchant Strength, Agility or Intellect then the fact that those three stats are far and away better than the others won't matter as much. They already put the exact same amounts of these stats on every item piece of the same level anyway because they can't condone any variation in their values, why allow gems to vary their values? If you couldn't it would mean that items with gems slots would be more customizable instead of just more powerful. They could also change the way they do itemization to make the primary stats more expensive in item points to reflect their power.

I would actually like to see a mix of these approaches. Having one stat be five times as good as another is a real mess for the game. To be honest, this problem seems pretty easy to predict and I will be interested to see exactly how bad things are in the beta and how bad they are when cataclysm actually ships. The biggest problem may ultimately be that if this is not addressed from the outset it will be almost impossible to address it mid-expansion and we might go through and entire expansion with nearly every class having only one stat that matters.

On the bright side, I guess that would solve the tank damage scaling problems.

Wednesday, May 5, 2010

Addendum: Hasted Channels

While they seem to be unsure about the issue, a developer recently said that hasted channels may or may not work the way hasted DoTs will. That is, they may add extra ticks to channeled spells to leave the duration of the channel unaltered by the haste. This is far more bizarre than their plans for hasted DoTs.

While there are some important differences in the game between a channeled spell and an ordinary spell that resolves at the end of its cast time, they interact with haste in exactly the same way. Fireball takes three seconds to cast, Mind Flay takes three seconds to cast. Put 100% haste on them and Fireball takes 1.5 seconds, Mind Flay takes 1.5 seconds. You cast more spells when you have more haste.

Unlike DoTs, making channeled spells interact with haste in this way would not increase dps at all. You would spend just as much time channeling Mind Flay either way, and do the same amount of damage in that time. What it would do is increase the mana efficiency of channeled spells. Why would they want haste to increase the mana efficiencies of channeled spells and not of direct nukes?

Of course it would also do some other crazy things. If you had haste procs that affected a certain number of spells (imagine a Nevermelting Ice Crystal for haste instead of crit) then it would mean that extra haste would become more effective at increasing the damage of your channels than increasing the damage of your nukes. The equivalence of damage and nukes relies on the haste being the same for both. If you can do a big haste for one spell, it is more effective on a channel. This is because 100% haste on Mind Flay means casting Mind Flay at double speed for 3 seconds, while 100% haste on Fireball means casting Fireball for double speed for 1.5 seconds. Basically, if you only get to cast a certain number of hasted spells (as opposed to casting hasted spells for a certain amount of time, in which case it makes no difference) then you effectively get to increase the number of hasted spells by the amount of haste granted if you cast channeled spells instead of non-channeled ones.

This is totally unintuitive (haste doesn't make things faster, it makes them do more damage), makes bizarre distinctions meaningful (haste has a significantly different interaction with channels compared to direct nukes when applied to a number of casts but is essentially the same when applied over a duration), and in the end changes almost nothing about how we actually play the game except in strange edge cases. Going forward with this plan would be ridiculous.