Monday, January 31, 2011

Strange Vengeance Math

I've been wondering how Vengeance worked for a while. Since you don't actually get a buff it can be pretty hard to tell exactly what is going on. Courtesy of the rawr tank modeling discussion on Elistist Jerks, I have my answer, and it is stranger than I could have guessed.

Your maximum vengeance value is, as a blue post indicated not too long ago, your stamina plus 10% of your base health. It was 10% of your health before they increased the health per stamina to 14, but they didn't want to increase Vengeance when they did that. For reference, a tank in 359s might have a maximum Vengeance value around 13k.1

When you take damage, if you do not have Vengeance, you start a new stack of Vengeance which is 5% of the damage you took. Then, every 2 seconds Vengeance ticks and does one of two things:

If you took damage in those two seconds then your new Vengeance value is 95% of your previous value plus 5% of the damage you took.

If you did not take damage then your new value is your old value minus 10% of the largest that stack of Vengeance has ever been.

So if you are taking continuous damage over time at a given rate, your Vengeance stack will tend towards the amount of damage you are taking per two seconds. That is, if you are taking 10k damage per second then your Vengeance will tend towards 20k. Of course that will be way above your cap, so you will actually trend towards your cap quite strongly.

Of course tanks don't take damage continuously. If a boss is swinging once every 2 seconds, and hitting for your Vengeance cap then that will be good enough. The problem is that you will avoid some of those attacks, causing a 10% decay in Vengeance. That will leave you with a 11,700 Vengeance stack. If you get hit again before the next tick then your Vengeance will be 11,115 plus 5% of the hit amount, so you'll need a hit of at least 37,700 to top you back off on that tick, or, more generally, a hit equal to 2.9 times your maximum Vengeance stack. Two misses will mean you will almost certainly not be back up in one strike.

This means that in practice your actual Vengeance will increase non-linearly with Stamina. The more Stamina you have, the more your maximum stack is going to be and the harder a boss will have to hit you to recover the stack after a miss. Of course higher stamina will mean higher stacks, but more stamina means less value for each point of stamina. Against Maloriak, for example, the marginal AP value of Stamina decays from .88 to .71 as you move from 11k to 13k max Vengeance.2 If you could reach 17k Stamina then it would go into freefall, going from .5 to .3 to .02 value per stamina as you got from 18k to 21k. Then at 22k max Vengeance something strange happens: your marginal return is negative. At some point, more stamina means less average AP from Vengeance.

Why does this happen? Given the long period I ran my simulation over, at some point you will dodge infrequently enough that your Vengeance stack will approach Maloriak's average 2 second damage. This will be the high value for your Vengeance stack, and will define the amount of Vengeance you lose if you don't take damage during a 2 second window. The higher this is, the more AP you lose when you successfully dodge an attack. If you are not consistently hovering near your maximum Vengeance stack then you are not getting much benefit from having a higher maximum, but you are still going to end up suffering the drawback of the higher stack size.

If you had no maximum Vengeance against Maloriak hitting you all day, you would end up with around 14.4k Vengeance - at whatever level of armor my Maloriak tank had last time we beat him. But if your Maximum Vengeance stack is around 21k, you will get an average of 15.7k Vengeance. You start doing better than no maximum at around 16k maximum, which is probably attainable this expansion, though obviously not while Maloriak is still relevant.

I don't think there are any bosses currently where more stamina would mean less Vengeance at intended gear levels, but I think Ignacious would close if it weren't for the fact that he is only one phase of the fight and leaves quickly. He has all the right ingredients: short periods of intense damage followed by drawn out periods of mild damage with occasional times where he stops attacking to guarantee a large decay of Vengeance. I can certainly construct reasonable sounding bosses in my head where appropriate levels of Stamina would pass the point where more Stamina means less Vengeance.3

Is this terribly important? I'm not sure. Apparently modelling tools are advising bear druids to gem for stamina, and I know returns from Vengeance are a part of that. I'm sure that advanced modelling tools are using reasonable estimates of Vengeance return for stamina points, but estimates can be reasonable at best, they cannot be accurate in a broad since because there is too much variation based on fight mechanics. I wonder if this is really the best advice a bear can get.

This would be an easy problem to fix. The value of the 10% decays for taking no damage should be set at the last time you took Vengeance damage, not the maximum size your stack has reached. Alternatively, there could simply be a 5% of current value decay every 2 seconds as there is when you do get hit right now, and have it rapidly decay and fall off when you are out of combat. If they don't want it to be that strong then they could simply lower the maximum value.

The current system is just a little too bizarre. If it gave great results then I could accept some strange math, but it gives wacky results. Wacky processes that give wacky results should be replaced with simple processes that give acceptable results.


1. This is the cap for a tank who is gemming heavily for Stamina. Depending on tank class, this might not be a good idea, so that cap might be much lower.

2. Using Maloriak's physical attacks hitting you for a day. In reality he has breaks in his attacks and you'll have to rebuild your stack, so it will be lower than this, spiking temporarily higher during red phase.

3. Imagine a twins fight. We'll make them faceless because I like faceless. So one twin is the caster twin and the other is the beater twin. You put on tank on each. The caster stands and nukes his tank with shadow bolts every 2.6 seconds (2 seconds but he is susceptible to cast slows), but sometimes he does cone of death centred on a random raid member with a 3 second cast (3.9 seconds for real). Every 45-60 seconds one of the twins puts up a shield and channels a spell on the other that makes him do way more damage, increasing over time. You break through the shield and interrupt before the tank on that twin dies. There are adds, void zones, explosions or whatever to do raid damage and keep people reacting to things.

In this fight, when the caster twin is being powered up he would do lots of damage and stack his tank to maximum Vengeance. Then his relatively lighter damage that comes more than every 2 seconds (the healers spend a lot of time on the raid) coupled with taking 3.9 seconds off hitting the tank now and then, would allow the stack to suffer from significant decay. The higher the maximum, the more the tank would be losing, but the stack would never fall off to reset the highest value for that stack because the damage is coming in too consistently.

This sounds like an encounter that could exist. It is an unlikely but believable mix of ingredients that allows Stamina to decrease Vengeance stack size.

Friday, January 28, 2011

Improving Normal Dungeons

This week I've talked about a number of problems with dungeons, and suggested that many of the perceived problems with heroic dungeons are really caused by normal dungeons being marginalized. I've criticized the awarding of Justice Points of normal bosses as an ineffective solution to this problem because it only serves to further marginalize normal drops.

Now I am going to present some of my solutions to these problems. These are real solutions designed to create a space for normal dungeons. With each of these solutions, I present what I think is a realistic timeframe for when this could be implemented given the current state of the game. These changes are radical, but probably the best thing about WoW and about Blizzard is that radical changes are considered if they are good changes. I will be submitting all of these on the suggestion forum.1

Fully Separate Normal and Heroic Difficulties
I've talked about this before, but it really is part of the solution. The current model for progression is normal dungeon to heroic dungeon to normal raid to heroic raid. The ideal would be to have two separate streams, normal dungeon to normal raid and heroic dungeon to heroic raid, possibly with some heroic players running a few normal dungeons just after reaching maximum level to get some gear.

Whether you choose to approach normal or heroic difficulty should be a legitimate choice based on your preferred playstyle and time commitment. Normal difficulty should not be a stepping stone on the way of heroic. Currently every progression stage has to be two things: 1) challenging for those who will never reach beyond that stage and 2) a gear grind for those who don't think of that stage as an accomplishment. At the same time as many people are struggling with heroics, many other people are already bored of running them every day for their 70 Victory Points. Having these groups play together is a recipe for conflict.

In this model the gear from heroic dungeons should be equal level to the gear from normal raids and heroic dungeons should be harder than normal raids. Both normal dungeons and heroic dungeons should be beatable in quest gear with good play - good play being a relative term depending on the difficulty level. Raids would be tuned expecting players to be mostly geared in appropriate dungeon gear. Because normal raids and heroic dungeons would give the same gear, those who played normal dungeons would be ready to try heroic raids to see see if they wanted to switch streams. But this would be a function of practice leading to increased play skill, not a function of getting more gear.

Timeline: I thought it was important to bring this is as an ultimate goal, but it is completely unrealistic to implement this until the next expansion. Changes being made now should be seen, in part, as preparation for this new state.

Less Gear from Vendors
When the emblems system first premiered in Burning Crusade it seemed like a really great idea. We all know the frustration of waiting for those last one or two drops to complete our gear sets, so having a way that we acquire a few pieces of gear for certain with sufficient work seemed like a good remedy. But in Burning Crusade, gearing up for raids involved a lot of running heroics for drops. In Cataclysm faction vendors and Justice Points provide so much variety in gear that dungeon drops are becoming insignificant. Waiting for that one drop your need is even more frustrating because it was the only drop you needed in the first place. Sometimes I would wonder why dungeons even had cloth gloves in them at all.

The amount of slots that can be covered by reputation vendors should be reduced, and the rate at which points are earned should be reduced as well. Points should be a way to fill in for slots you just can't seem to get rather than the primary reason to run dungeons. Each difficulty setting should earn points only at their own difficulty. So heroic players earn Victory Points but not Justice Points. This would prevent the marginalization or normal difficulty gear.

Timeline: In Cataclysm this ship has already sailed at the gear cannot be taken back. It might be possible, though, to hybridize this concept in with the current one by not giving out Justice Points like candy in heroic dungeons as soon as the next major content patch.

Three Dungeon Difficulties
Normal difficulty dungeons are currently trying to fulfill two incompatible roles. First of all, they are the ends of the stories of each zone. The storylines presented through quests end in assaults on enemy stongholds that are the dungeons. For this reason, normal dungeons need to be extremely accessible to everyone, and they need to be accessible at the level appropriate to the zones. As a result, only a handful of those dungeons are appropriate for maximum level characters, and it creates the impression that normal difficulty is a throwaway to be played through once only.

I am proposing a third difficulty be added - Story mode - that is intended to be played through at a level appropriate to the zone the dungeon is in. This difficulty would replace the current normal difficulty (i.e. all current normal dungeons would instead be the "story" dungeons) and a new normal difficulty would be placed between heroic and story. Normal difficulty would be for maximum level characters, and would present many of the same challenges as heroic difficulty but be more forgiving in terms of damage outlay, dps checks and time to adapt to or avoid fatal attacks.

This would mean that those who want to play at a lower difficulty level would have a full range of dungeons to experience, including dungeons like Deadmines and Shadowfang Keep that have been updated for maximum level. Normal dungeons could be something that people choose to play with their play time, rather than being seen as just glorified group quests.

Timeline: I honestly think this should be implemented in 4.1 or 4.2. That may sound completely crazy because of the amount of work involved, but I think it's pretty easy to overestimate how much work it would really be. These new dungeons would be exactly the same as heroic dungeons, but with some numbers dialed down. Because they are being tuned to be easier, they do not need to be tuned as tightly, so applying a ratio across the board for damage and health would probably suffice as a first pass with a second added here or there to abilities that you have to react to to stay alive. With future dungeons more care could be put in to balance properly based on lessons learned from the first batch. The infrastructure already exists for difficulty levels, so it is hard to believe that adding another would be a more massive project than is normally taken on in major content patches.

Dungeon Fragment Finder
Limited play time is a definite hindrance to experiencing content, especially if normal dungeons were a little harder than they are now. With the current dungeon lockout system, however, it would be possible to implement a system where someone could essentially save their progress through a dungeon2 and then later look for a group to finish it. I necessary people who were not saved at the same point could be brought in, since people are generally pretty happy to join a dungeon run already in progress to reach the end boss faster. A good progress saving system would also be helpful for guilds running dungeons together who play for limited amounts of time.

Timeline: Patch 4.2 or 4.3. I'm sure there are some technical challenges to implementing this system, and it is a lower priority suggestion. It would be a valuable upgrade to the dungeon finder, though, to make dungeons more playable with limited play time.

Normal Difficulty Achievements
Achievements add a lot of replayability to heroic dungeons. When a boss has been mastered, there is something else to try for. Similar achievements should exist in normal to add the same amount of replayability.

Timeline: I think this could be added any time after the third difficulty, but if another difficulty is not added then achievements would need to wait for the next expansion. Unfortunately normal dungeons are just too easy to overgear to make these achievements even feel like achievements, and most of the normal difficulty dungeons are not for maximum level characters.

Less Loot Inflation
Reduce a "tier" of gear to only 7 or 8 item levels. This would help with scaling problems from the beginning to the end of the expansion and also reduce the role that overgearing plays in defeating content.

Timeline: I'd really like to see this change in 4.1, but I'm not holding my breath.

Better Information About What Went Wrong
This is actually something the developers have mentioned they would like to do and I wholeheartedly support it for normal difficulty. If you are killed by Rajh's Inferno Leap, getting a message telling you that that was what killed you, and that in order to survive it you have to get away from where he is landing, would be a great addition. I know that currently this is primarily intended for heroic difficulty, but in the long run if normal and heroic difficulty were separate then I don't think this would be necessary for heroic difficulty. Playing heroic would indicate a level of comfort with figuring things out on your own.

Timeline: I think we'll see something like this in patch 4.1 or 4.2.

Suppose all of these things had been done for Cataclysm, what state would we be in now? Imagine that by the time you completed all of your solo play you had access to one piece of 333 loot from a quest chain and two pieces from getting to revered with all the factions. The rest of your loot is 325 from quests. You choose, based on your own preference, whether you want to play normal or heroic difficulty and start looking for dungeons of that difficulty.

In terms of execution and play skill requirements - not in absolute terms - heroic bosses are close to the same challenge level as the easier normal raid bosses currently are. A heroic boss dungeon boss would put up nearly as much of a fight for a newly geared 85 as Halfus, Double Dragon, Magmaw or Omnitron would right now for a full 346 geared group. If you choose normal you find that those bosses are a little harder than current normal bosses, expecially in the sense of being less forgiving of mistakes, but that they have significant amounts of instruction in terms of raid warnings plus indications of what went wrong if you lose. By succeeding in these dungeons you would be able to gradually upgrade all your gear to 333 for normal or 341 for heroic, giving you in the neighborhood of a 10-12% increase in suvivability, healing and damage. Introductory raid bosses would be toned down in their difficulty from where they are now by a noticeable margin (remember that if you were playing heroic you would not do the bosses on normal first), and final bosses could be either a little easier or stay the same given appropriate adjustments for lower gear numbers.

Casual players would feel that they were accomplishing something by playing normal dungeons with their limited play time. "Hardcore Casual" players would get to approach heroic dungeons with like-minded players. Servers would have a small number of heroic guilds and a larger number of normal guilds. Raid accessibility would be increased - a goal of developers since Wrath - without watering down content for those who like the hardest challenges.

This would allow a wider array of players to access dungeon and raid content, generate less friction by putting players with very different play styles and goals together less frequently, and give players more space to set their own goals instead of being forced down the one progression path.

If you think these changes sound insane or counterproductive, I'd love to hear about it how terrible I am. If you think that they sound awesome, help to spread the word. Blizzard is very responsive to customer feedback, and the more a good idea gets echoed in the player base, the more likely the are to notice it.


1. So far I have quite a remarkable record of my suggestions being implemented, which probably means that I generally have made suggestions that the developers were already planning on doing anyway. I hope I'm behind the curve on this stuff too and they already have it in the works.

2. This is already essentially possible with the lockout system by just extending your last instance, but the interface is somewhat hidden and not user friendly.

Thursday, January 27, 2011

Justice for Normals

The developers have heard the complaints about heroic dungeons and they have responded. In patch 4.0.6 the rate at which you will be able to earn Justice Points by doing normal dungeons will be greatly increased. Not only are they doubling the Justice Points earned by doing a daily random to 140, they are giving 30 Justice Points per boss for Halls of Origination, Lost City of Tol'vir and Grim Batol.

The idea, it seems, is to make normal dungeons more rewarding and to help players enter heroic dungeons with more gear. But how does this goal measure up to reality?

First of all, let's consider how good a source of Justice Points normal dungeons will actually be. Bosses being worth 30 versus 70 Justice Points seems like a big decrease, but they also die a lot faster. Heroic Anraphet has 4.9M health while normal Anraphet has only 2M. If you wanted to farm Justice Points and had the choice of fighting either of those bosses repeatedly, you would choose normal for the simple reason that 30 divided by 70 is greater than 2 divided by 4.9.

It's not that simple, of course, but heroic trash has a lot more health too. The ratio isn't consistent, but aside from the fairly common 309,960 / 387,450 mobs, heroic trash usually has about 75% more health than normal trash and bosses have between 2.3 and 2.8 times as much health. Adding up the total health of normal versus heroic Halls of Origination gives the slight edge to heroic in terms of Justice Points per mob health, and of course the time to run between mobs is the same on both difficulties.

So heroic is the clear winner for Justice Points assuming that you have the same amount of difficulty with both. If your healer drinks more on heroic, if your dps die more, if you ever wipe or stop to discuss strategy, if you mark mobs for CCing when you could be using AoE abilities to do more damage, then normal is a better place for you to earn Justice Points. The number of slots you can expect to get 346 items in from normal dungeons and heroics will reach 10 of 16. If this does prepare you for heroics, it also leaves you with few things left to upgrade when you get there.

I think all of this runs the risk of substantially mischaracterizing the people who are going to play normal difficulty dungeons on an ongoing basis. I seem to be talking as if everyone who plays the game is only interested in a frantic gear grind. I know that this is not the case, but I think that how the frantic gear grind works has ripple effects that will affect everyone. So let's divide the audience for normal dungeons in two categories: 1) Those who are playing normal dungeons because they like to play WoW and that is what is accessible to them in the play time they have; and 2) Those who are playing normal dungeons only as a stepping stone to gear up for heroics.

The first group is a group that doesn't get talked about a lot. If you only log in for 30-90 minutes at a time (maybe a little more sometimes on weekends) and you only play WoW two to three times a week then it is a pretty safe bet that you also don't spend a lot of your time talking about WoW on forums, reading WoW blogs or otherwise making your presence known in the WoW community. The dungeon finder was a huge boon to people who play in this category because they could run dungeons in the limited time they actually play the game. The dungeon finder can still work for this group of players, but only for normal, not for heroic instances. A typical random heroic takes too long to complete, and may be too difficult.

The second group has a playstyle that is going to be antithetical to the first. If I felt like the best way to upgrade my chest piece was to run normal Halls of Origination 11 times then I would do it, but I certainly wouldn't want to take my time doing it, explain fights, or possibly even exchange pleasantries.

By making normal dungeons a viable source of gear for people who are having trouble with heroic dungeons, these changes will force these two groups to play together, which will be less fun for everyone. They will be a disservice to the first group because they diminish the rewards of the dungeon in favour of rewards to be purchased from a vendor. They will be a disservice to the second group because they give people the impression that the way to solve their gameplay problems is by grinding better gear. If Corla's friends are ascending, then at 346 chest is not going to help you, so people are going to end up grinding up gear in normal dungeons that have become boring for lack of challenge and then find that they are still having the same problems with heroics they were having before.

Tomorrow I'll be concluding my Normal Dungeon Manifesto with a list of suggestions that would improve normal dungeons by making normal dungeons seem like a worthwhile goal unto themselves and by responding to the needs of players who run normal dungeons.

Wednesday, January 26, 2011

Bad Habits

Cataclysm was a rough time for many players who had developed very bad dungeon habits in Wrath of the Lich King. The group up and AoE model for trash, often taking more than one pull at a time, didn't translate to the new dungeons and the new healing model. This adjustment to CCing, paying attention to enemy abilities, and otherwise playing well was never going to last, however, and for the most part my guildmates have already returned to killing everything at once in heroic dungeons because 359 gear allows that.

But if 359 gear enables the group up and AoE strategy in heroics, what level of gear is required to enable that strategy in normal?

Let's look at the abilities from normal and heroic trash. Here I'm intentionally choosing some of the most deadly trash mobs available in level 85 dungeons. These mobs aren't necessarily that dangerous if handled correctly, but will easily kill individuals or groups if their abilities are ignored.

Ascendant Waterlashers have wiped more than one group I've been in because their front cone has been facing the group. This cone lasts 4 seconds and deals damage every 0.25 seconds. On normal it averages 1500 damage per tick, on heroic it averages 4000 damage per tick.

Temple Shadowlancers channel Pact of Darkness doing high damage in melee range, and often killing off melee dps who do not interrupt it while also putting a strain on tank healing. Pact is channeled and deals damage every 1 second. On normal the average damage is 4500, on heroic it is 18000.

The penalty for not dealing with an ability correctly on normal is in the range of 27000 damage to a single character or up to 96000 damage spread around the group. On heroic it is instead over 108000 to an individual or upwards of 256000 damage divided among the group.

Not all heroic versus normal mobs have such extreme differences. In many cases, the damage difference seems to be around 50%, but spells that you are supposed to avoid or play around are often much higher. All of this brings us to the fact that if you walk into normal with some 333 gear and some 346 gear then there is little reason to CC trash or react to abilities. Failing to interrupt a spell that an enemy is casting doesn't mean very much. Facing the enemies the right way doesn't mean very much. Targeting the right enemy first or burning down a dangerous enemy doesn't mean very much. The gear you can get from soloing is adequate to group up and AoE in normal instances.

Equally important are the boss abilities. On normal, General Umbriss' Blitz deals 30k, on heroic is deals 100k. On normal, Erudax's Shadow Gale deals 3k on the first tick, increasing to 30k by the last one for a total of 165k, on heroic it deals 10k on the first tick, increasing to 100k for a total of 550k. On normal failure to dispel Anhuur's Divine Reckoning means 15k damage, on heroic it means 45k. It is hard to directly compare bosses outside of level 85 dungeons because you encounter them at different levels, but the numbers are still telling. The rocks High Priestess Azil throws on normal deal only 35k damage, which is survivable for a level 82 character. On heroic they deal 400k damage which is not survivable for anyone. Ozruk's Shatter is similar, dealing only 30k on normal - a hit you would expect your tank to simply take - and 150k on heroic - an instant fatality to new geared heroic characters. With the new healing system, I often see healers saying, "I can't heal stupid." In normal dungeons, you can.

When the developers say that players who are having trouble with heroic difficulty may want to try normal difficulty to learn the fights, they are ignoring the fact that you can't really learn many of these fights on normal. Learning only happens when there is a motivation for it. When your tank gets killed by Pistol Barrage the first time you face Godfrey, you ask questions like, "Can you move out of that ability or should you use a cooldown? Will he move to keep facing you when he's casting or can you escape without taking any damage at all?" If we imagined a level 85 normal version of Shadowfang Keep, the tank would be taking manageable damage while standing in the barrage. Those questions may never get asked, and while the healer would walk out of the fight saying that it seemed like a lot of damage, a win would be quite possible while ignoring all the mechanics. If you never have to ask questions like that to win, then you never learn to ask questions like that when you aren't winning. What is going wrong in heroics becomes a mystery - it seems like you just can't heal enough.

You don't learn Altairus' upwind/downwind ability when you can heal the fight with the haste penalty. You don't learn Throngus' Phalanx ability when there is no penalty for standing in front of him and your healer isn't in danger of running out of mana. In some cases key abilities don't even exist on normal. Normal Ozruk doesn't cast Paralyze, normal Beauty pulls separately from her pups, normal Setesh doesn't have attackable portals.

The longer you stay in normal dungeons, the more gear you get, the easier they become. But the bad habits that people form in normal dungeons are going to do more harm than the gear they collect can do good. This is why we see healers complaining about getting kicked from groups after the dps make unhealable mistakes. After enough experience in normals, "unhealable" leaves your head as a concept.

Tomorrow I'm going to look at the solution that the developers are implementing in 4.0.6 to make normal dungeons more attractive - increased Justice Point rewards - and what effects we should realistically expect from that change.

Tuesday, January 25, 2011

Normal Dungeons Manifesto

My first excursion into a heroic dungeon was somewhat challenging. It was Stonecore, which is undoubtedly one of the harder heroics, and Corborus gave us a bit of a run for our money. Of course it was because we were playing terribly, getting hit by his charge while he was underground, but it left me with a good feeling that heroics really were harder now, and that encounter mechanics could not be ignored in favour of overgearing. As much as I like the new heroic design, there are clearly many who don't, and they regularly show up on forums to talk about how heroics are no longer fun for them. My gut reaction is to defend heroics because I don't want them to change or to be made easier, but that adversarial relationship is just masking the real problem. It's not that heroic dungeons aren't fun, it's that normal dungeons aren't fun.

This post is the first in a four part series on what went wrong with dungeons in Cataclysm from the perspective of a casual player. Today I'll be talking about how players can acquire gear and why normal dungeons are out of place in that progression. Tomorrow will be about how the current system of dungeons creates a dichotomy between normal and heroic dungeons where the best way to beat a heroic is by playing well and the best way to beat a normal is by having better gear. Thursday I will look at the role of Justice Points and what we can expect from the change to the way normal-difficulty players can earn Justice Points in 4.0.6. On Friday I'll present some real solutions to some of these problems.

Before I get to the topic I really want to discuss, and that is where gear comes from for non-heroic players, I want to talk about a few of the common concerns that are echoed around the forums. I don't want to dismiss 45 minute queue times for dps, the length of time it takes to clear heroic dungeons, or the challenges of doing difficult content with people you don't know through the dungeon finder, because I think all of these are legitimate issues for some people. But I think that too many people are trying to find solutions for problems that they perceive with heroic dungeons when they should be trying to find solutions to problems with normal dungeons. Almost everyone wants content to be tuned so that it feels like a challenge but it is ultimately surmountable. Unfortunately, a single dungeon cannot be that to all people - it has to run the gamut from unbeatably hard to boringly simple depending on the players that enter it. In focusing all efforts on finding ways to fix heroics, we are trying to make them into something they can't be - content for everyone. There are two levels of difficulty for dungeons, and if we work on fixing the problems with normal difficulty dungeons then instead of having one level of difficulty that is meant to be something for everyone, we could have two meaningful levels of difficulty and allow people to play the one they have the most fun with.

So why aren't normal dungeons working? One problem is definitely that there are only three of them for level 85 characters. This doesn't give enough variety for players who want to player normals dungeons. Even if all eight heroic dungeons were playable at level 85, though, they still wouldn't see nearly enough play.

A casual player, or any player, has most likely leveled up by going through the various zones and doing the quests. While some people have have leveled mostly from archaeology, this is certainly the exception rather than the norm. Having done all those quests means having gathered quest rewards, and having a decent amount of reputation will all of the new factions in Cataclysm. Questing alone will get you to, or pretty close to, Revered with the various factions. That means a lot of gear, including many 346 pieces, are readily available to players who have done nothing but solo content. Let's take a look at what a cloth-wearer1 could have in terms of gear from solo content alone.

You can get 346 items in the following slots: Back, Chest, Finger, Finger, Weapon (2-handed)
You can get 333 items in the following slots: Feet, Hands, Head, Legs, Waist

That means you are settling for less than 333's for your Neck, Shoulder, Wrist, Trinkets and Wand. Of course there are dungeon quests that give you 333's for Shoulder and Wrist, so one run through each of two dungeons guarantees those upgrades. That leaves you with only 4 gear slots that do not have 333 gear in them by the time you have run two normal instances even if you don't get any drops. Some drops in normal instances may be side-grades for some of your slots, but you already have many 346 items making many of those drops completely useless.

In fact, with a little bit of effort and an appropriate profession, you can hit the 329 ilvl requirement for heroics without actually doing any normal dungeons. Normal dungeons are a stepping stone that most players don't even need to step on if they want to avoid it.

Given this is it no wonder that players are complaining about heroics instead of enjoying normal dungeons. What is there to enjoy? Normal difficulty dungeons were virtually designed as throw-aways to be run once at most. Furthermore, many of the upgrades you could in theory get in normal dungeons have Justice Point vendor items in the same slot, making any upgrades that can be found feel temporary at best.

Some blue posts have suggested to people that if they find heroics too hard, they may want to consider gearing up a little more. This advice from blue posters would be helpful if normal dungeons were actually a place to gear up, but for the most part they are not. While it is possible that running a few normal dungeons could increase average item level by 1 or 2, either by filling out a rep to revered or by actually finding a drop, that kind of gear upgrade is not really what people who are struggling with heroics need.

Wrath of the Lich King gave a large part of - probably most of - the player base the impression that for dungeons (if not for raids) these difficulties should really be called easy and normal. There was no hard mode for Wrath dungeons, but there was a very trivial mode that people spent little or no time in. Part of making heroics harder, should have been trying to reverse this impression. Normal mode for dungeons should be what it is for raids: the difficulty that the majority of people find success in. Unfortunately by making the rewards of normal difficulty dungeons easy to match or beat through quests and solo play, the developers trivialized the rewards of normal dungeons and left players little reason to want to spend time in them.

Tomorrow I'll talk about why running dungeons on normal difficulty could be making things harder, rather than easier, for players who want to graduate from one to the other.


1. I am assuming that they can make use of either Spirit or Hit, so I guess I am excluding shadow priests.

Monday, January 24, 2011

Why Not the Right Number?

Prayer of Healing is getting a little bit of rebalancing in Patch 4.0.6. It's effect is being reduced by 15%, but Divine Aegis is going to have an additional effect on it. From the looks of it they thought Prayer of Healing was too powerful for holy priests, maybe a little less so for discipline priests, but they wanted to make Crit rating having a bigger effect on it for discipline.

So they changed Divine Aegis so that you get twice the effect out of it when you get a critical on Prayer of Healing. As things stand now, Prayer of Healing may hit for around 7k and put up a 2.8k shield with around 14 mastery. On a critical it will heal for 10.5k and put up a 4.2k shield instead.

With the changes that same Prayer will only hit for around 6k and put up a 2.4k shield. On a critical, however, it will hit for 9k and put up a 7.2k shield. The non-critical heal is 15% lower, but the critical heal is 10% higher. This change makes your Prayer of Healing worse if you have less than 52% crit and better if you have more. We aren't really hoping to see 52% this expansion, so this change is definitely a reduction.

The amount of crit you need for this to be a good thing is sensitive to the amount of mastery you have, but at all levels of crit and mastery we will have throughout the expansion, this is going to be a slight reduction in Prayer of Healing power. Since the glyph does not account for the Divine Aegis shield, it is an even greater reduction if you use the glyph.

I already had the suspicion that Prayer of Healing was too powerful, so I'm not surprised by this. They are also providing significant boosts to non-Prayer of Healing heals from Discipline priests, most notably Power Word: Shield, but also Penance and to some extent all other heals through the Grace change. I'm not unhappy with the changes, aside from worrying that the Power Word: Shield buff is too big and will result in an extreme buff/nerf cycle.

What is really bothering me, though, is exactly how much extra shield you get when you crit with Prayer of Healing. Say you have enough mastery to get 40% of your heal value as a Divine Aegis. In that case, if you critical with a spell your critical does 1.5 times the normal amount and then gets an extra 40% in the form of a shield for 2.1 times the base heal. With a prayer of healing, your normal heal + shield is 1.4 times the base healing. The critical amount is 1.5 time the base heal with a shield equal to 80% of that for 2.7 times the base heal. That means it is 1.93 times the non-critical value. Prayer crits for 93% extra and other heals crit for 110% extra. The relationship varies with the amount of mastery you have, with Prayer falling further behind in the value of crit the more mastery you have.

I don't understand why they wouldn't just pick the value that actually made Prayer of Healing crits the same as the crits on other spells. The current implementation just seems so inelegant.

Tuesday, January 18, 2011

Heroic Dungeon "Nerfs"

I am a little bit late to the party in terms of weighing in the heroic dungeon changes. Many other bloggers have made the point that they like hard things and don't want everything to get dumbed down to Wrath levels. At least one post says almost the same thing I'm about to say, but like everyone else on the internet, I assume I am a unique snowflake and have my own valuable take on things.

I don't want to go through the changes line by line the way Righteous Orbs did, since I think if you want that you can just read that post. I want to say a few general things about changes, though, and why I like the thrust of these changes in general.

It's important when a change comes to not get too caught up in how things are now. Changes are changes because there was a way it was before the change. If there isn't they are new things rather than changes. But we tend to get too caught up in how things were and don't pay enough attention to how things should have been in the first place. Changes that change things so that they are how they should have been in the first place are good changes, even if we kind of liked something about the bad state things were in.

Many of the "nerfs" to these encounters seem a lot more like bug fixes. While they may not always deal with literal bugs in the code, they are correcting what we may call bugs in the design.

First of all, there are many changes to the visual effects of boss abilities to make it easier to distinguish what they are doing. Being able to distinguish what the boss is actually doing does make encounters easier, but it is also how things should be in the first place. A boss that dropped invisible fires would be harder than a boss that dropped visible fires. A boss that dropped fires that appeared to have a radius of 4 yards but really had a radius of 10 yards would be harder than a boss who dropped fires that looked as big as they are. A boss that requires you to hit your "G" key (regardless of what you have it bound to) at 1:30 into the fight to avoid dying without giving any indication that this is what you had to do would be harder than a boss that did not have this requirement or that put up a raid warning telling you you had to do it. But in each case the former is a pretty bad way to make a boss hard1. If the boss is going to drop fire, the visual effect should line up with the game effect of the fire. If you want to make it harder, do so by making the fire do more damage, or giving me less time to get out of it, not by not letting me know where it is or what I am supposed to do.

Second there are changes to bosses that seemed overly sensitive to group composition. I agree with the principle but not the form of many of these changes. There should not be a boss that requires two beast CCs to beat, or one that requires interrupts more than once every 10 seconds, or one that requires two snare effects. Bosses should be beatable by a tank, a healer and three dps in ilvl 329 gear with good play.2 The specifics of how Beauty, Ashbury and Erudax were nerfed didn't deal with the issue appropriately, though. Beauty went from being nearly impossible with only one beast CC and fine with two to being fine with one beast CC and really, really easy with two. A better solution would be to make the pups a lot less threatening, give them less health, and make Beauty herself more powerful. Ashbury could have simply healed for less with Mend Rotten Flesh, perhaps 10% instead of 25% of his health. That way interrupts would have been good but not required. The Faceless Corruptors are now still pretty hard to deal with if they are not slowed but even easier if they are slowed. They should have had their movement reduced more dramatically and been given an immunity to, or at least a severe resistance to anything that would slow them. Maybe reduce the duration of slowing effects on them by 75% or 80%.

Third there are changes that don't really make the encounter easier, but people call them nerfs anyway. Foe Reaper 5000 does take a long time because you are down a dps. Making it a little shorter doesn't really make the fight much easier to master, just a little less boring after you have mastered it. Slabhide spending little time on the ground was a key factor in allowing us to beat him more than a minute and a half after our tank died. If he'd spent more time on the ground then shadow priest tanking would have been unsuccessful. Grand Vizier Ertan may have less health, but that's just to compensate for the fact that you'll actually have to pay attention to his mechanics now.

Of course the patch has undeniable nerfs in it, but I agree with most of them. Setesh seemed significantly harder than his compatriots on the second floor of the Halls of Origin for PuGs. A little more time to avoid Corborus and Ozruk's sudden death attacks is welcome as those are real PuG killers. A reduction in damage from the Naz'jar Tempest Witches sounds like a great idea, as does the reduction in damage from trash in Throne of the Tides.

And when looking at the "nerfs" it's important to notice that some bosses were made harder, and rightly so. Ptah, Asaad, Walden and Anraphet were all real pushovers - I doubt the changes to them will even go far enough towards making them harder, but we'll see.

So what don't I like? In addition to the specifics of the composition sensitive boss changes I mentioned above, there are a couple of boss nerfs that I simply don't understand. Specifically, I don't understand why Valiona was nerfed in the Drahga Shadowburner encounter. No one is supposed to be getting hit by that ability anyway, and the challenge in that encounter is dealing with adds properly, so the change makes little difference. I also don't see why Throngus' impale shouldn't still be a stun. I know people don't like their healer getting impaled, but if you play that fight right he just doesn't do that much damage. I really don't understand the Ozumat nerf in the Neptulon encounter either - but maybe it's just that Prayer of Healing is very good for that part of the fight.

There are also a few things that I find it hard to believe they didn't nerf. I feel comfortable calling Springvale the hardest boss in any heroic, and I think he could stand to have his melee damage reduced, even if only a little. Also, they reduced the damage of Gilgoblin Hunter's Poisoned Spear attack "slightly." Rather than "slightly" that should probably say, "by half," and there should be a similar note regarding the aquamages and their tsunami. Having a pack with eight enemies, each of which has a 40k attack is a little bit much. Generally how that pull goes is the tank pops a cooldown while running in, we start AoEing and just before we kill them four aquamages tsunami at once killing the tank and any melee, then we clean them up. Chained tanking CDs for the duration shouldn't be the norm for trash pulls, and, more importantly, should be sufficient to avoid one-shots if you can actually pull it off.

I also wouldn't have minded seeing a few other changes:

Blackrock Caverns
Corla in BRC should have her haste aura removed and her attack speed and damage increased so that she does roughly double the damage she does now in the end. This would make the fight easier when her friends evolve and not trivial if they don't.

Halls of Origination
Veil of Sky should be made more powerful and undispelable while Astral Rain should be reduced if it is left for the third ability on Isiset - the damage seems fine at both the first and second thirds of the fight, but appears to be multiplied by about 10 if you leave it as her last ability. The rain does nearly twice as much, hits all five people instead of just two, and lasts seemingly forever. Simply affecting everyone with no damage or duration increase would have been fine.

Shadowfang Keep
Baron Silverlane could have used a buff of some kind. Lord Godfrey should have the damage of his cursed bullets dot reduced and his other damage increased to make the fight less dependent on whether you have a curse remover. I don't mean that a curse remover shouldn't be helpful, but he should be beatable in reasonable gear without one in the group.

Throne of the Tides
Mindbender Ghur'sha should make your allies more threatening when he controls them, take control more often, and do something when he's flying around by himself.


1. I could actually imagine invisible fires having a place in some boss fight at some time, but it doesn't seem likely, and the other two examples are right out.

2. Giving paladins tanks an interrupt will also help with this.

Monday, January 17, 2011

Q&A

Swaggart Asked:

Is bubble spam feasible with the current regen numbers in-game? For example, can you actually cast PW:S 81 times on Maloriak in T11 gear?

Whether or not bubble spam is feasible depends on what is meant by "spam." In the post in question I divided the total healing that I did on an actual kill of various bosses by the approximate size of a Power Word: Shield from patch 4.0.6 to get the number of shields I would cast to contribute the same amount to the fight. This is a naive comparison to see whether shielding is an affordable way to heal, but is pretty far off the shield spam that discipline priests did in Wrath.

Maloriak has a six minute enrage times, and will almost always take pretty close to six minutes to win since for the foreseeable future we will need to wait until after the second green phase to put him under 25%. I'm almost in tier 11 now with an equipped item level of 356, so I'll just use my stats as a decent approximation. With all raid buffs up over six minutes I would have the following: 124k mana pool, 196k from in combat mana regen, 45k from replenishment, 11k from hymn of hope, 10k from a potion and 37k from shadowfiend. I'll talk about Rapture a little later. That's 423k mana. The cost of shield with Mental Agility is going to be 4626, so 81 shields will cost 375k. Clearly 81 shields in six minutes is doable with a lot of mana to spare.

Actually spamming shields, in contrast, is totally out of the question. Thanks to Borrowed Time I could in theory currently cast a shield every 1.17 seconds. That would be 307 shields in six minutes, costing 1.4M mana. That would probably be possible to 11 restoration shamans in my raid, but not under any plausible circumstances.

In reality using just shields to heal would almost certainly be wrong in a 10-player raid. Not casting Penance - especially now that we know it is also getting buffed - seems like a big mistake. It seems, however, that going from 4.0.3 to 4.0.6 a discipline priest will be able to match their old output using only shield and still have mana left over to do healing after that.

Does the inflated absorbency of PW:S mean this it will take a little better timing to get a rapture proc every 12s?

I do think that the larger size of the shield will make it harder to predict when your Rapture procs will come in some circumstances. Without predictability, you will have the choice of casting early and possibly "wasting" the chance at a proc or casting late and accepting that your procs per minute will be closer to 4 than to 5.

That being said, bosses hit pretty hard and a lot of the time their AoE damage will break shields. Halfus will do multiple Furious Roars followed by a Shadow Nova. Valiona does Blackout on the ground and Twilight Meteor in the air. On Maloriak breaking a Flash Freeze or taking a single hit of Scorching Blast will break shields. Magmaw does spew and spit at the same time. Shields won't be broken when a boss merely glances in your direction, but they will still be broken by randomly targeted attacks and most AoEs.

I think that actually getting a Rapture proc close to every 12 seconds is very hard. I used to get them close to every 14 seconds in Wrath, and these days I only cast shield every 15 seconds at most. But in 4.0.6 you will no longer need a Rapture proc to make shield worth casting. It's going to be one of your most mana efficient heals - and will totally blow competitors out of the water in terms of throughput - and discipline priests will go back to casting it a lot. It will be the heal of first resort instead of a weird mana return ability. Casting more shields should net more Rapture ticks, even if they take a little longer to break. I think we'll see Raptures between 13 and 14 seconds apart fairly consistently just because we'll actually be casting shield a lot. In the six minute fight above, that would be about 191k mana - over 500 mana per second and a great use of three talent points.

Do you think that Disc would fit in better with the current design philosophy if PW:S had a lower benefit from Mastery, but DA and maybe even PW:B got (better) use of the stat?

I honestly don't know what they could do to discipline mastery right now to fix it. I seems like they just need to come up with a completely new idea. It's too bad they already gave a healing-creates-shields mastery to paladins, since that would have made sense for discipline priests. Like other masteries, they are presumably want to keep it all down to a single number, so I doubt that affecting shields one way and aegis another way is in the cards, but it might be a good way to handle things. Unfortunately, the bonus to aegis would have to be over the moon to actually make the stat good for single target heals. If paladins get 1.25% of their heal to a shield every time, then it seems like discipline priests would need about an 11% boost to the effectiveness of DA per point of mastery since it is a percentage boost to 45% of the normal heal size less than 25% of the time.

I could brainstorm up some ideas for mastery for discipline priests, but it would just be a wishlist or trying to pretend I can comshow backseat develop the game I'm sure things will look different a year from now, but for 4.0.6 I think discipline priests should just enjoy being overpowered.

Thursday, January 13, 2011

A Few More Comments on Mastery

This post is basically part two of yesterday's post so reading that first is probably a good idea if you haven't.

I finished off talking about the problems with a mastery rating that only affects one spell by saying that I hope the developers realize that the one spell mastery is the real problem and change that, rather than trying to change the power level of either the mastery or of Power Word: Shield when the problem rears its head.

Using Spell A and Spell B I just want to talk about why you can't solve the problem by tinkering with the numbers, but it has to be solved by redoing the mastery. In my example, Spell A is a two second cast for 20 damage and Spell B is a three second cast for 25. 25% of the time Spell A procs an instant cast of Spell B. I pointed out that if Mastery only affects Spell B and it is balanced with haste then it will get out of hand and eventually you'll just spam Spell B and mastery will be overpowered.

So lets compare how the numbers would actually work out. I was using the example of having 10 mastery which would give a 58.8% bonus to Spell B. First I want to show why this is "balanced" with haste. One mastery takes 1.4 times as much rating as 1% haste, so we should compare 10 mastery with 14% haste. If you do the rotation you are "supposed" to and cast Spell A until you get a proc, then use your proc, your basic rotation does 105 damage in 9.5 seconds, for around 11.1 dps. Put 14% haste you instead get 105 damage in 8.33 seconds, for 12.6 dps. If you get 10 mastery, increasing the damage of Spell B to around 40 then you get 120 damage in 9.5 seconds, again 12.6 dps.

The problem, as we noted, is that the mastery threw out the balance between the two spells and created a different rotation. The alternate rotation is worse with low levels of mastery and better with high levels. Once you cross over the threshold where it gets better, mastery becomes a dominant stat. At 10 mastery the spam Spell B rotation gives 13.3 dps, so it's 5.6% better than the rotation you are supposed to do.

What if we nerf mastery so that the rotation of spamming Spell B doesn't get out of hand? Easy enough to do. If reduce mastery by a little bit then we could find an equilibrium so that 14% haste and 10 mastery give the same dps. The problem is that this equilibrium would exist for those exact numbers only. If we wanted to reduce mastery so that 10 mastery gave 12.6 dps at most, then we'd need mastery to give 5.12% to Spell B per point.

But then what about a few tiers later when we have 15 mastery or 21% haste? The mastery option lets us spam Spell B for 14.7 dps while the haste option does only 13.4 dps. We nerfed the mastery massively but it ends up almost 10% better later on.

We could nerf mastery to make sure that spamming Spell B never became a real option, but that would mean making sure it never got above 36.9 damage per cast. If we know the most mastery people will be able to get in the expansion is 15 then we can set the value of mastery at 3.17%. So then how does 10 mastery compare to 14% haste? 10 mastery would give only 11.9 dps. It would result in overall dps being 5.5% lower than the haste option. Haste rating would be about 87% better than mastery for increasing your dps.

If instead of reducing mastery we reduce the spell, we'll either just recreate the problem in a more extreme fashion or we'll create a situation where the best way to play is to stack haste and just spam Spell A, ignoring the procs.

The numbers just don't work out. Even at a specific gear level, we can reforge, regem, re-enchant and get new items. Discipline priest mastery only really makes sense at extremely high crit levels. If we were critting 60% of the time then it could be affecting our non-shield spells more consistently. There would still be a significant imbalance in how it affected the spells, but it would be much less pronounced. On the other hand, if we could crit that much, why would we not just reforge our mastery into haste and stop casting Shield altogether in favour of spells that can actually benefit from crits?

Mastery for discipline priests, as it is, is unbalanced. In future tiers - or possibly right now in 25-player raids - it will go from being unbalanced to degenerate, and the only way they will be able to stop its degeneracy, without a redesign, is by making it useless.

Wednesday, January 12, 2011

One Spell Mastery

Suppose you were playing a dps class with two spells: Spell A takes 2 seconds to cast and does 20 damage. Spell B takes 3 seconds to cast and does 25 damage. Sometimes, however, Spell A gives you a proc that makes your next spell B instant. This would make for a pretty obvious rotation.

If you got 1% haste, you would do 1% more dps. If you got 1% crit you would do 1% more dps - but since critical strike rating takes 40% more rating to get 1% they might give you a talent that makes your crits do more to even out crit and haste. What should your mastery be? If your mastery was that you do 1.4% more damage per point, then it would be balanced with crit. Suppose, though, they thought that was boring, so they made your mastery only affect Spell B.

Spell B composes 23.8% of your damage, so if a point of mastery is supposed to increase your damage by 1.4%, it would have to increase the damage of Spell B by 5.88%. Haste, critical strike rating and mastery are all balanced.

There is a problem with this situation, however. Suppose you get 10 mastery. Then Spell B does 58.8% more damage. That means it hits for nearly 40 a shot for a three second cast. Spell A still only does 20 for a two second cast. Your highest dps rotation is just to cast Spell B instead of casting Spell A and waiting for procs. Furthermore, mastery is now *way* more powerful than crit or haste. You should ditch everything you can to stack mastery and blast away with your single spell rotation. The higher the tier your gear gets to, the more stacking mastery and spamming Spell B will outstrip a normal build.

Of course this is pretty easy to understand from a dps perspective. It's no wonder that no dps class has a mastery that affects only one of their spells. They didn't seem to realize, however, that this can have the same unbalancing effect on healing.

Discipline priest mastery doesn't literally only affect one spell, but it comes pretty close. Here is the percentage increase given by 100 mastery rating on my spells: Power Word: Shield - 1.01%; Prayer of Healing - 0.3%; All others - between 0.1% and 0.14%. Mastery is more than 10 times as effective at increasing Power Word: Shield than it is at increasing Atonement healing from Smite and more than three times as effective on Power Word: Shield as it is on Prayer of Healing - the only other spell that consistently affects. Compare this to how 100 intellect affects the throughput of my spells: Smite - 1.19%; All others - pretty close to 0.67%. I've talked about Smite scaling before, so I'll ignore that for now.

If I want to make my spells more powerful, I should obviously gem Intellect. Since it is also a better mana stat than Spirit, gemming intellect is the clear choice. If, however, I want to make Power Word: Shield and only Power Word: Shield more powerful, I should gem Mastery instead. Could I use the same theory applied in the dps example above, do nothing but stack mastery, and cast only Power Word: Shield?

Lets say I manage to get 90% of my items to have mastery rating on them reforge in the other slots. That would be 1974 mastery rating. Also I put mastery gems in all sockets (we'll go with two in a staff, chest, head and legs and one in other socket-prone slots) for another 560 mastery. I can get 115 from enchants and two 321 mastery trinkets. This is a total of 3291 mastery.

Of course my other stats would suffer badly. Sacrificing Intellect and Spirit for Mastery would really hurt my mana. But using the numbers from the PTR for Power Word: Shield, it would boost that one spell to 7.72 healing per mana and 25.7k healing per second. Right now my highest efficiency spell aside from Penance is Smite at 6.27 and my highest throughput spell is a virtual tie between Renew and Flash Heal at 17.4k.

Talking about the viability of this plan is hard to do. Let's start by asking an easy question: could I put out similar total healing numbers by stacking mastery and only casting Power Word: Shield? Each of my shields would be around 35.7k healing and shielding. Dividing the healing I did on a few bosses by that, I get the number of shields that I would have to cast to contribute the same amount of healing to the fight that I did on our last win.

Here they are: Halfus - 45.5 over 5:55; Double Dragon (Valiona and Theralion) - 120 over 8:38 (Not the cleanest kill ever); Ascendant Council - 71 over 6:47; Magmaw - 84 over 5:26; Omnitron - 89 over 6:24; Maloriak - 81 over 6:01.

So the highest rate I'd need to cast shields to keep up with my normal healing is one every 3.9 seconds, and a typical rate is one every 4.4 seconds. Would I have the mana to keep that up for nearly 9 minutes? I would actually have enough mana to keep it up forever. Assuming I can get a Rapture proc every 14 seconds, I would be very slowly gaining mana while casting at this rate.

So the final question is whether that distribution of healing would be viable for winning fights. You'll notice I didn't include some fights in my list, such as Conclave and Chimaeron. Healing those fights with shield alone would certainly not be feasible unless that shield was very big indeed. But casting nothing but shields is not necessarily a bad healing distribution for typical fights. Magmaw will always eat through shields with damage would have to be healed otherwise. Shielding Arcanotron's target and/or shielding everyone during Magmatron is fine. Shielding everyone before Maloriak does his fire breath would be a huge help. Shielding everyone on Magmaw and and Double Dragon would work out. Shielding everyone would work great for the important part of Council.

I don't think this is actually the right way to play a Discipline priest - not for 10-player raiding anyway. It may be precisely the right way to play a Discipline priest for 25-player raiding, and it will almost certainly be the right way to play for 25-player raiding in the final tier of gear where the difference between stacking mastery and having more reasonable stats will be more pronounced. Of course by then the crazy scaling on Smite will make Atonement very overpowered as well. Maybe bringing two discipline priests to raids will be a good idea after all, but more likely we'll see some hideous nerfs around at tier 2 or 3. I just hope that they realize that the problem is that they gave Discipline priests a mastery that only affects on spell, and no amount of tinkering with the power of that spell will really correct things.

Tuesday, January 11, 2011

Rethinking Veiled Shadows

In my build I was taking Veiled Shadows because more Shadowfiend seemed more good. In fact, I think I was mostly taking it as a holdover from a time when the beta version of the talent was reducing the duration by 1 minute per point instead of 30 seconds per point. A shadowfiend every three minutes instead of every five is certainly worth two points. I am now less than convinced, however, that a four minute shadowfiend is worth the investment.

First of all, let's look at how much mana the talent gives us with a simple once-every-four minutes instead of once-every-five minutes analysis. Shadowfiend restores 30% of your maximum mana. In my case, that's 36000. 36000 mana every five minutes is 120 mana per second. 36000 mana every four minutes is 150 mana per second. So you get 30 mana per second for the two talent points. I haven't made a post about benchmarking talents yet, but this is clearly not that strong.

The real advantage of reducing the cooldown is not that it gives you 30 more mana per second if you cast it on the cooldown every time. It's that it allows you to get two shadowfiends in a fight where you would normally get one. If a fight is five minutes and you get to cast a second shadowfiend because of the talent, then that's 36000 more mana for the fight, or 120 mana per second over the fight. The question is, will Veiled Shadows actually do that for us?

I'll make this easy on myself to get a ballpark figure, and then talk about the complexities. First of all, I can ignore mana regeneration from any source that regenerates mana proportional to the length of the fight. So ignore spirit regen, Rapture, the Mp5 buff and Replenishment. When I talk about how much mana I am spending, I will be talking about how much I am spending above that continuous mana regen. I'm also going to assume I'm spending mana at a continuous rate.

If I get to Hymn of Hope once during the fight, then that will restore 9.2% of our base mana1 and a mana potion is around another 8.3% of my max mana. So I have around 114.5% of my base mana pool to work with. Just because I'd rather work out of 100%, I'll instead say that shadowfiend restores 26.2% of my augmented mana pool.

I wouldn't want to cast shadowfiend as soon as the fight starts because most of the mana would be wasted. The earliest you would want to cast shadowfiend is:



Where mana_spent is measured in percentage of augmented mana per second.

Veiled Shadows will give me an extra shadowfiend cast is the length of the fight minus t if between 240 and 300 seconds. We probably want to ignore at least those cases where the second shadowfiend doesn't last its full duration, so look at cases between 255 and 315 seconds instead of those between 240 and 300. Of course getting an extra cast isn't the only thing to consider. If Veiled Shadows is giving me an extra cast, am I actually going to use that mana? On the flip side of this question, am I going to have any mana left when it comes time to cast shadowfiend again? If this model has me running out of mana at the three minute mark, then it isn't a model of a winnable fight and Veiled Shadows is irrelevant.

So the amount of mana I have left when it's time to cast my second shadowfiend is:


And if we substitute in the above equation for t this simplifies to:


Which is pretty easy to work with. Our conditions above were that the second shadowfiend matters, so the amount of mana we have left should be less than the amount of mana we are expecting to spend in the rest of the fight, and also that we should actually have enough mana to get to the second shadowfiend without running out, so the amount of mana we have left is greater than or equal to zero.

Simplifying these two requirements, we get the following:

and

Substituting one into the other, we arrive at the conclusion that the length of the fight must be greater than 303 seconds in order for the second shadowfiend to be helping us. Basically in a five minute fight there is no way for the second shadowfiend to work out, and there are a lot of five minute fights. In the event that you are spending almost exactly 0.416% of your augmented mana per second, this talent would probably be very good, and maybe even necessary, but possibly not enough to make winning the fight feasible with your current mana.

Of course going over 303 seconds is no guarantee that the talent is working for you. We need it to not only be the case that a four minute shadowfiend is working for us, but also that a five minute shadowfiend would not be.  Using similar analysis for mana expenditure per second, we can see that starting at around 379 seconds, the five minute shadowfiend starts being viable for a second use.  The four minute shadowfiend won't be viable for a third use until at least ten minutes.   Furthermore, the windows to take advantage of the extra shadowfiend are quite small.  In order for the talent to be helping, you need to be casting your first shadowfiend between approximately 72 and 80 seconds into the fight. If you are casting it sooner - and not wasting it - then you are burning through your mana so fast that you have no chance of making it to the end of the fight. If you are casting it later then you are using mana at a slower pace and don't need the extra shadowfiend cast anyway.

The shape of damage on a given fight might alter these things. Veiled shadows might be useful on a fight like Halfus where you spend a lot of mana early on, then have a big lull.  Conveniently, Halfus is also a six minute fight, which is where Veiled Shadows really has an opportunity to come into play.  Use shadowfiend when you are still fighting the drakes and you might both get full effect out of it and have plenty of mana to make it to your next cast, which you might need because you are forced to heal up after the roars. This kind of high damage at the start, high damage at the end, lull in the middle fight, however, is pretty rare, and it is also a pushover that you don't need a special talent spec for.

Overall, I think I am just not hitting the window to make this talent work for me. The question is, where to put the other two points. For now I think I'll stick them back in Surge of Light, but with patch 4.0.6, both Improved Renew and Soul Warding present themselves as reasonable alternatives.


1. Hymn of hope gives 2% mana for each of four ticks. However, it also increases maximum mana by 15% for 12 seconds. Assuming you get all four ticks, that means that the second through fourth tick give 2.3% of your normal maximum mana, since your maximum mana will be higher, so it gives 8.9% of maximum mana. But also you have increased maximum mana for a total of 18 seconds from when the first tick hits you to 12 seconds after the last tick. That means 18 seconds of getting 15% more replenishment, which is 0.1% mana per second, which means 0.27% more mana maximum mana.

Monday, January 10, 2011

Inspiration versus Surge of Light

Inspiration versus Surge of Light was one of my least certain decisions when I was putting together a talent spec. When I swapped to Smite from non-Smite, it seemed like Surge would be better given that Smite does not proc Inspiration but does proc Surge.

Well, I have some real data now, so I can actually see how much physical damage was dealt to people in the 15 seconds following me landing critical heals on them. I'm working from our World of Logs reports, so it's a little cumbersome, but foolishly I haven't been running my own logs up to this point.

The result for Magmaw is that Inspiration would have prevented around 90k damage, and on Omnitron it would have prevented 139k. On both encounters I healed around the same amount, approximately 3.15M. So as a percentage of total healing, it was 2.85% on Magmaw and about 4.19% on Omnitron.

I calculated the expected number of Surge procs given the coming patch that will allow Flash Heal and Greater Heal to proc it. Surge did not fare to well, with only 3.5 procs on Magmaw and 2.9 on Omnitron. Even allowing that the free Flash Heal will be able to crit *and* will be substantially more likely to benefit from Grace, this is substantially less than the damage reduced by Inspiration.

Of course these numbers are pretty far from telling the whole story, given the very different ways that the two talents function. Surge gives a heal that takes my time to cast but that I get to personally direct. Inspiration is sort of a blanket over the raid, randomly reducing bits of damage. But I think in the intangibles category, Inspiration has a bit of an edge for a two reasons.

First, Inspiration does not take time to cast to extra "healing".

Second, while I would normally be in favour of benefits that I get to personally direct rather than benefits that are handed out at random, when the personally directed benefits are so heavily dependent on procs, they become very random. I can't make the best use of an instant cast, free Flash Heal if I can only use it three times a fight and if my windows to use it are only 10 seconds long. If those windows don't correspond to the best time to use it, then I am having to make decisions to direct it without getting the benefit of making decisions. In this circumstance, I'd rather have it just be handed out at random.

If the way in which the benefit is handed out is preferable and it's a bigger quantity of health over the fight then the choice is pretty clear. I'll be moving those points over in my build.

Friday, January 7, 2011

I Feel Like a Paladin

Lots to digest in the patch 4.0.6 notes, but here's the big one:

Power Word: Shield now scales from 87% of spell power, up from 41.8%. Base absorb increased by 110%, from 3906 to 8136. Now costs 25% of base mana, up from 19% of base mana.

Now I think it was obvious that I was clamouring for some kind of buff to Power Word: Shield for while. That being said, I'm not sure that giving Discipline Priests a 29k on-demand instant heal is the solution to that problem. Compared to Greater Heal it would have 67% more throughput and 20% more efficiency without a Rapture proc. As before, Rapture procs are larger than the cost of the shield. Power Word: Shield would have higher throughput than a four target Prayer of Healing.

I guess the question is, what is Power Word: Shield supposed to be doing? There are three states is seems to be able to occupy: 1) High throughput and efficiency plus all the benefits of being a shield instead of a heal so cast it all the time instead of regular heals; 2) Mediocre throughput and low efficiency without a Rapture proc so cast it once every 15 seconds on the tank like clockwork; and 3) Useless.

The ground I think we'd like to see Power Word: Shield occupy is where we actually cast it on someone who is at low health when they are in danger of dying so that another heal can land, or on someone who is about to take massive damage to make sure they live through it, but that we don't cast it all the time as a staple heal.

I really see only one way to accomplish that, and I don't think anyone is going to like me for suggesting it: lower the duration. If Power Word: Shield only lasted 8 seconds then you wouldn't go around spamming it on people as a heal because it would fall off without doing anything too much of the time. If someone was low and you were worried about them dying before they got a heal, it would work just as well as it does now for that purpose - but would have an extra "hidden" cost that it might go to waste. If someone was about to take a big hit, 8 seconds would provide lots of time.

I'm now getting very nervous about what's going to happen to discipline priests. The more buffed you get, the bigger the nerf is going to be. Normally this kind of swing is reserved for paladins, who tend to either be overpowered or underpowered while priests have traditionally been in the middle. We may soon return to the days where discipline priests were a strange kind of healer: you are crazy if you don't bring one, and almost equally crazy if you bring two.

Thursday, January 6, 2011

Melee Hate

With the recent comment from Paragon on their Ascendant Council kill and the need to sit melee classes in favour of ranged, I expect a lot of people are talking about the problems with melee dps and how so many fights are designed to punish us for being near bosses or being near one another.

I'm actually not that worried about Paragon stacking ranged on a fight, because they would obviously stack whatever it took. If it came down to it they would bring 8-10 of a single class that was ideal for an encounter. But it isn't that one fight that is causing the problem, it is nearly every fight. The problem is not that developers have been going out of their way to make things hard on melee, it's that their entire toolbox makes things harder on melee.

To quote paragon: "Here's to hoping next tier of raiding won't favor ranged by design. Maybe even go wild and give some incentive to bring in melee, too."

What might incentive to bring melee look like? To be less ambitious, what might simply not favouring ranged look like?

Any ability that requires people to spread out, clump together, be far from the boss, move to a specific spot, or really to be anywhere at all favours ranged. Any ability that requires a switch of targets favours ranged. Any ability that makes an area dangerous or puts fire on the ground favours ranged.

There were fifty bosses in Wrath of the Lich King. Of those bosses 17 actively punished you for bringing melee, or for bringing more than one or two melee; 18 had mechanics that noticeably favoured ranged with positional requirements; 5 were pretty much the same for melee and ranged provided that you did not bring a group very heavily weighted towards melee; leaving 9 where there was no substantial difference and only one where it was better to bring melee. Even on the fight where I think melee were definitely preferred - Anub'Arak - that was because of "free" cleaves, a mechanic that has gone away.

The reason why ranged are generally better than melee is simple: being able to attack at range is better than not being able to. There is no arguing that point. A warrior must be 0-5 yards away from an enemy to attack, a mage must be 0-40. There is no compensation for the warrior for this disadvantage.

So what can be done to fix this? What encounter mechanics can make it so that we'll want to bring more melee to raids? I'm not sure there are any. If a mechanic punishes you for being 30 yards from the boss then the ranged will just stand next to the boss - this would be a hunter punishing mechanic, not a melee favouring mechanic. In order to avoid that, they have to put a reason why someone has to be 30 yards from the boss. In that case it has become a melee punishing mechanic. With mechanics like that a raid of 2 tanks, 3 healers, a rogues, a death knight, an enhancement shaman, a warrior and a feral druid just can't win - a pretty severe punishment. Best case scenario for such a raid is that a healer is performing the job that requires range and dodging whatever they have to dodge.

The only thing melee have going for them right now is that melee classes consistently have interrupts on 10 second cooldowns. But even then, shaman have a 6 second ranged interrupt, so this is not a clear advantage and it isn't a reason to favour melee.

If there are no mechanics that can favour melee, then what can be done to make it so melee is not worse than range? Should they do more damage? Unfortunately this would create a host of new problems. Suppose melee did 10% more damage than ranged and that melee hate remained the way it is. Take a boss like Kel'Thuzad who creates very difficult positioning for melee classes, and limits the number of melee you can bring before you run into problems. A 10-player raid only has 5 dps, so a 10% buff to one of them is nearly a 2% buff to the raid. If the boss has a six minute enrage timer than 2% is a difference of 7.2 seconds. So how much health does the boss get to make the enrage timer threatening? If they tune it so that you are supposed to have three melee and deal with the positioning then thats like cutting 22 seconds off the time that a group of ranged dps has to kill the boss. If they tune it so you don't need to bring melee then a group with good melee dps can beat the fight in much lower gear than was intended.

They could make melee do more damage, and rely on the melee punishing effects and the time they spend running around to cut it back to the same as ranged, but then melee classes do the same damage and require more skill to play.

I don't have a solution to this one. The current situation is that ranged can always be substituted in for melee and the reverse is not true. I don't think that is a healthy state for the game, but I really don't see what to do about it, since any advantage given to melee would just tend to create a problem in the other direction. The one option I can think of is to give all ranged classes s minimum range similar to that of hunters, but the consequences of that for PvP and solo play would be disastrous.

Wednesday, January 5, 2011

Mana Tide Totem

I did a little bit of work on mana over the past few days1 and there are a few important conclusions that distinguish Cataclysm from Wrath. One is that Replenishment is no longer the ultra-important buff that it used to be. Of course you don't need a spreadsheet to figure this out, the calculation is pretty simple. If you have 100k mana and it gives you 1% of your mana per 10 seconds, then it is giving you 100 mana every second. In a five minute fight that's 30k mana. We all want 30k more mana to spend over five minutes, but we can manage without it more easily than we can manage without some other buffs. Of course you may have more than 100k mana - and if you are a discipline priest who is raiding you really ought to have well over 100k mana, but replenishment is now only about 60% to 100% better than the Mp5 buff given by Blessing of Might and Mana Stream Totam2. It's a very powerful but probably not absolutely necessary buff.

Of course getting replenishment isn't that hard. Six different classes can supply it with one of their specs, so most raids should be able to access it. A significantly more powerful mana buff, however, is available to exactly one spec of one class.

Mana Tide totem increases Spirit by 350% for 12 seconds every three minutes. The 12 second duration can be increased by 40% by a talent to 16.8 seconds. That gives it an uptime of around 9.3%. Getting 350% more of a stat 9.3% of the time is pretty similar to getting 32.6% more of that stat all the time, if the gains can be averaged out. Of course this is actually the worst cast scenario for the totem. You get a Mana Tide every three minutes at least, but in a five minute fight, you probably get two, which is one every two and a half minutes. This would increase Mana Tide to a 39.2% buff to spirit. It's all a little odd given that they said there would be no buffs to spirit in the game.

So how much mana does this actually provide? In a five minute fight, using my current gear, it provides me with a little over 50.5k mana. Druids, paladins and shamans all get a little less than that, in the 45k-48k range. Holy priests get around 84k mana from it - making mana tide about a 25% boost to the mana available to a Holy Priest over the five minutes.

Of course this kind of bonus dramatically alters the value of your stats. If you have a shaman to drop the totem for you then spirit is about 60% better than intellect of paladins and shamans, nearly on par with intellect for druids and discipline priests and a whopping 100% better for holy priests at generating mana. Without it, those numbers decline dramatically with spirit being only 20% better than intellect for paladins and shamans and similarly worse for the others. This kind of difference should significantly affect gemming choices. As a discipline priest, I'm much more likely to match a socket bonus with a blue gem if the spirit is as good as Int at giving me mana than if it is only two-thirds.

And of course mana tide totem stacks. I don't think it literally stacks, but they can be staggered to increase the uptime. More restoration shamans means more mana.

Currently the most powerful mana buff in the game by far is available from one spec of one class only, and stacks if you have more of that spec of that class. This buff situation does not seem appropriate.


1. I built myself a little Mana Calculator to play with. It's not very user friendly and the estimates for the mana gains for Water Shield are pulled out of a hat - two issues which I may correct in the future if I am interested in doing so.

2. In the first tier of Wrath raiding replenishment was about two and a half times as good as the Mp5 Buff. By the end it was four to five times as good.