First, however, I must lay out the history of Lunacy and define the style of guild it is. This is, afterall, the context in which I give advice, make decisions, and formulate opinions.

What Lunacy Is, and What It’s Not

Lunacy is a top 100 guild in the U.S. Our kill of M’uru placed 40th in the U.S., while our defeat of Kil’jaeden slipped us down to 64th. With this in mind, we can be considered hardcore. However, we are not a top 20 guild. We are not currently on the same level as any of the guilds on the front page of WoWProgress. Nor did we strive to achieve such a feat during The Burning Crusade.

I’m not about to pretend we’re something we’re not. However, to have killed Entropius (M’uru) well before his first nerf, and to have defeated Kil’jaeden two months before 3.0.2 is no small feat. This puts us in a relatively small class, at the 99.6th percentile of ranked guilds on WoWProgress.

Furthermore, having survived both M’uru and Kil’jaeden as a guild, whom people so fondly refer to as “the guild killers” is also an accomplishment I think is a testament to our guild and our ideals. We value the skill in our members and are constantly evaluating the level at which we play. If we cannot succeed or efficiently progress with the members we have, either we must improve or change our members. In this regard, we can be considered hardcore.

At the same time, however, I am not about to set a raid schedule of five, six or seven days a week. I value the lives of our members, as well as my own. It is simply impossible in this day and age for a person to raid on such a rigorous schedule and still manage to hold some semblance of a normal life. For this reason, we can perhaps be considered casual.

Hence, I always like to refer to Lunacy as a guild with a casual schedule that can accommodate raiders who approach their play in a hardcore manner. We are a “hardcore-casual” guild.

A Brief History of Lunacy

Lunacy did not exist during 1.x. Well, it did, but it was an entirely different guild than it was during The Burning Crusade.

During 1.x, I raided with various guilds. These included Exigence, Black Wolf Mercenaries and Project Mayhem on Proudmoore, as well as Zero Tolerance on Azgalor. My times with Exigence, Project Mayhem and Zero Tolerance were very short. And while my time with Black Wolf was extensive, I could no longer raid with them regularly once they had transitioned to a schedule that was exclusively Australian. After my stint with Project Mayhem, I evaluated my guild options and decided Proudmoore had nothing to offer me. One guild didn’t use voice communication, which I thought was imperative to success during certain encounters. One guild had a prominent officer that ninja’d loot from the raid of a close friend of mine and gloated about it on IRC. Another only raided weekends. And the last option I had was another Australian guild. With all of these guilds ruled out, I decided I would create my own guild at the start of TBC, and spent my time raiding casually with Black Wolf, PUGing BWL, running a PUG MC, and playing the TBC beta.

At first, I was going to transfer servers and start the guild elsewhere. I thought the Australian-American split between Proudmoore’s population was too problematic. However, I realized it would be difficult to establish a guild as a fresh face, and my co-GM convinced me to stay by offering to help lead the guild. This was when the modern version of Lunacy was born. I originally did not want to adopt the same name as my first guild, but Silver was a part of the original incarnation and insisted on the name. So I buckled and agreed, even though they are completely different guilds.

From there, we began recruiting. At the launch of TBC, we had merely five people who intended on raiding with us full-time. After a couple weeks, we had only eight capped members and raided Karazhan with a couple friends filling the last slots. From there, we slowly built our guild from the ground up, running heroics nearly every day to trial potential members and invite those we felt were adequate in meeting the guild’s standards.

Initially, our goal for the guild was set relatively low. The general idea was to create a guild that had the potential to break down raiding walls efficiently and clear all of the bosses in any given content cycle. This is something Proudmoore failed to do in Naxxramas, never defeating Gothik in Naxx-60 before the release of TBC. It was my intention to fill this hole and challenge other guilds by creating a guild that would compete on a level higher than Proudmoore was seeing in the waning days of 1.x.

From Hydross on, Lunacy achieved this goal. The only two bosses we did not achieve sever firsts on in tier five content and above were Lurker and Illidan. Illidan was a major blow to the guild, no doubt. However, it did not dent our morale and instilled in us a desire to rise again to number one in the Sunwell Plateau. Also, it was easy for us to take heart in losing the server first on Illidan, because we only lost by 30 minutes and actually managed to do something no one else in the world had done at the time. We were the first guild in the world to achieve a progression kill (that is, a “guild first”) with a paladin tanking Illidan.

Losing Black Temple to Renaissance did bring us back down to reality, however. Following the loss, we heightened our recruitment standards and vowed to approach Sunwell with an aggression that would elevate us to an entirely different level. If someone’s DPS still needed work, we simply erred on the side of caution and denied them entry into the guild. If they couldn’t deal with various situational abilities, they didn’t make the cut. However, I will admit mistake in being loathe to replace, cut or demote some of the existing members whose performance wasn’t up to our new standard. In this regard, we were slowly evolving, but there were some problems that persisted because we were still a developing guild.

As time passed, we became more concerned with our residual problems. After stalling for some time on Felmyst, Silver and I decided that we would sit anyone we felt wasn’t up to the task at hand for specific encounters. If someone was likely to conflag the raid on the Twins, they would only come in if we absolutely needed them. If someone’s DPS was too poor, we’d put in someone else on M’uru. This new caution and standard led us to place extremely well on M’uru (40th in the U.S.).

However, we weren’t done making refinements and our problems caught up with us. With the midsummer months bringing forth a great amount of attrition, anyone who hadn’t defeated Kil’jaeden faced waning membership and a limited recruitment pool. For this reason, it took us two months to defeat Kil’jaeden. Admittedly, this was perhaps also because we faced some internal problems outside the scope of membership. This experience provided me with perhaps the most perspective on just what I needed to do with the guild to make it successful in the future. For one, recruitment needed to be more aggressive. For another, I needed to act on not only performance problems, but also attitude problems. You can’t build a good raid or succeed if one or two people are souring the mood. And you can’t build a good raid if people don’t adhere to your guild’s philosophies. It is not enough to have people who show up everyday. You need to have people who exemplify what you expect from your raiders. You need to fill holes in your raid. And you need to hammer home your guild’s philosophies.

With that said, we now have a raid that I believe is even stronger than it was during our first kill of Entropius. I have a great amount of confidence in the foundation we’ve built for Wrath, which is something we obviously didn’t have during TBC when we were entirely new. That said, there are challenges ahead in transitioning to a new expansion. But I am not ready to discuss my plans for the future, as they are still being evaluated and decisions are still being made.

Needless to say, the evolution of Lunacy has been astounding and I have a great foundation on which to further build and improve the guild during Wrath.

What to Expect from This Guide

This series of posts is being written for people who hope to create or lead their own hardcore raiding guilds, and for those who find themselves thrust into the leadership ranks of such guilds. This compendium will offer advice and provide contextual examples for the creation, management and improvement of guilds that want to raid on a hardcore level, but maintain an atmosphere that breaks the mold from most other hardcore raiding guilds.

The series will cover the creation, building and refinement of a hardcore raiding guild. It will provide contextual examples from my own experiences, as well as references to the guilds who have ideas I sometimes like to emulate. It will begin by detailing the situations that can lead to the creation of a guild, followed by ideas on how to develop interest in the guild, how to evaluate players interested in joining, how to improve the efficiency and skill of the raid after the foundation has been set, and how to make any renovations necessary to heighten the success of the guild.

Recruitment particularly has required more attention than previously. During our time working on Kil’jaeden, and shortly after the kill, my guild suffered a massive amount of attrition:

• 1 holy/disc priest: to focus on life. I do not wish to explicitly say why, to protect his privacy, but it’s a very good reason.
• 1 resto druid: to become a helicopter pilot.
• 1 resto shaman: as a commitment to his new wife. He is available on occasion, however.
• 1 holy paladin: to join the navy.
• 1 mage: to pursue a relationship and focus on life. He is available in emergencies, however.
• 1 mage: to focus on work and life. She is available in emergencies, however.
• 1 arms/prot warrior: to take college courses that occur during raids.
• 1 rogue: mostly because of time differences, I think.
• 1 shadow priest: for reasons I’ll leave unsaid.
• There’s also a couple who disappeared without really saying anything, though one of them explained why after-the-fact.

This means we lost eleven people total leading up to our defeat of Kil’jaeden. Of course, we did gain a couple people in this time, as well. But the predicament resulted in dropping our core raiding numbers down to a dangerously low level. And applications weren’t exactly rolling in, since most of the available people out there applied to guilds that were 6 of 6 in Sunwell, instead of 5 of 6. And those that did apply to us weren’t qualified or didn’t pass the trial. We also had problems where people would occasionally miss raids for summer-related activities like family barbecues, birthdays, short vacations, etc. Now that we have Kil’jaeden down, however, applications are rolling in at a decent pace.

At a certain point, when recruitment pools are limited, attrition makes it extremely difficult to set up a raid that can succeed in downing bosses. Even nerfed, M’uru still requires a good balance of DPS: enough melee such that each door can be taken care of efficiently, but enough ranged such that gravity balls during Entropius are less impactful if they spawn near and leash onto the melee. The fight is certainly a joke compared to the pre-nerfed version we killed back in June, but it is by no means easy to repeat every week. There were nights where we would be playing with only one experienced resto shaman and two formerly enhance shamans playing resto to cover for our losses. Or, worse yet, no experienced resto shamans, one formerly enhancement shaman playing resto, and someone playing an alt shaman. The situation certainly came to an extreme in this regard.

Luckily my guild in now in a position to recover from our losses, however. Already, I have a resto shaman and a resto druid queued for trial, we’ve been trialing a holy/disc priest, and we’ve picked up a mage. Furthermore, people have been expressing interest in the guild everyday. So I’ve found myself standing deep in leadership work with merely recruitment. It does beg the question, however: where do you find the time to lead? I’ve been so swamped with reading applications and interviewing potential applicants I’m simply overwhelmed!

Where Do You Find the Time to Lead?

It doesn’t take a lot of effort to simply lead a raid or a guild. Anyone can invite people and run raids on a superficial level. However, if you want to lead a successful raid, there’s a rather long list of daunting tasks to perform. Personally, after the departure of my guild’s previous “administrative assistant,” I was taking on all of these duties:

• Outside of raids:
  • Assessing what classes we need, and making sure recruitment posts were created and updated on World of Raids, MMO Champion, WoWProgress, and the realm forum.
  • Reading applications and determining whether or not a person is worth interviewing for a position in the raid.
  • Interviewing applicants over Ventrilo if they warrant a further look.
  • Deciding if an applicant passes the interview and should trial with us.
  • Distributing epic gems.
  • Distributing sunmotes.
  • Evaluating attitude problems that might have occurred during or outside of raids and making decisions about them outside of raids.
  • Updating the web site.
• During raids:
  • Determining who is in and out for each encounter.
  • Assessing some of our trials and recruits.
  • Determining whether trials or recruits make the cut.
  • Taking interest in loot, and putting in my vote as part of the loot council.
  • Keeping people focused and calm.
  • Organizing the healers.
  • Sometimes organizing the DPS if it’s needed.

This comes between personal pursuits of blogging, training to run half-marathons (and eventually marathons), attending classes, homework, work, spending time with family or friends, and simply enjoying WoW for what it is. The latter point I think is important. What is the point of playing a game if you simply can’t enjoy it? Luckily for me, part of the enjoyment I get is in leading itself. I can sit back and laugh about the workload. However, I also like to take some time to just relax and mindlessly grind honor or levels at times.

It’s crucial to note the list of my duties doesn’t consider those my co-GM covers, which include helping with raid pacing, providing input as the other part of our loot council, paying for the guild’s web site and Ventrilo servers, distributing flask tokens, selling HoD’s, leading the tanks during raids, making sure we have extra consumables for the tanks in case they run out or didn’t have the time to farm for that day’s raid, and recruiting people through EJ’s benefactor bar. He also helps discuss and make some decisions outside of raids. It also doesn’t cover guild-specific tasks other guilds might have, such as maintenance of a DKP system.

To be blunt, it is impossible for one person to find the time to do everything on their own, unless they are retired or have a sponsorship that allows them to perform these tasks all day, everyday. Otherwise, it requires real life sacrifices many people aren’t willing to make–myself included.

Delegating to Find the Time

It is important for any good leader that doesn’t find all the time they need to delegate various tasks to others in the guild.

With the recent departure of one of our prominent officer figures, I’ve had to delegate some of the tasks to help make the raid more efficient. I am still in the process of refining this delegation, however. One of our rogues has been helping find potential applicants, and this has been a large part of the increase in applications we’ve received. As a highly respected member of the WoW community (the top rogue theorycrafter on EJ), people are keen in his advertisement for the guild. One of our shadow priests has been helping us track raids for several months. Furthermore, one of our warriors has now taken on the task of taking interest in loot when it drops and gathering information about people’s loot history, so we can make decisions while I simultaneously facilitate swaps for the next encounter after a kill.

That said, I have to talk with my co-GM about when we give them access to the officer channel and such. And I am still trying to decide who would be a good candidate for gem distribution. That way my focus will be primarily on evaluating and interviewing applicants or recruits, facilitating group composition for each encounter, leading the healers, keeping people focused, distributing sunmotes, maintaining a good atmosphere, and updating the web site. This is a manageable amount of work for me, whereas the three weeks before this one has largely been unmanageable.

How Much Delegation Is Too Much?

One of the things I’ve discovered over several years of playing MMO’s is that having too many people in various leadership, officer or officer-like positions actually has the potential to create a system that is too mired in “bureaucratic” stagnation. This creates a contrasting problem against the need to delegate tasks so you actually have the time to lead a successful raiding guild and prevent yourself from burning out.

Even in a co-GM system, where we primarily make the decisions, I’ve found there are certain topics too sensitive to act on before consulting my co-GM. For example, I am hesitant to kick or demote members without his consent. And what I’ve found is that he is extremely reticent to kick anyone. This means I often let issues of these kind drop and compromise by simply abstaining from putting people in if their attitudes are problematic. But his reluctance has admittedly prevented bad decisions. So it has slowed down the decision-making process, but it’s made it so both of us are not too mired in guild-based work.

I will say I think it’s important to have at least one other person who can fill in when the traditional leaders are absent or indisposed. However, I also think it’s imperative for a leader to exhibit restraint in just how many people they tap to help them. If minor stagnation occurs with merely two people leading a guild, imagine how much worse the problem would be if five officers all got together to hold a vote on various issues. The process would simply aggravate situations that require swift decisions. And sometimes you need quick choices and drastic actions to prevent further problems.

The article assumes that the bloom portion of Lifebloom from being offensively dispelled is the same as the bloom that occurs naturally from one that naturally expires.
–Xinhuan in a comment on the original article.

This is a valid concern expressed about the original tests. So further testing had to be conducted to strengthen the argument and provide more empirical evidence suggesting the end heal generates absolutely no threat, regardless of how the “bloom” is triggered. So an addendum must be made. Such is the way of the scientific and dialectical methods.

To make certain the purged bloom is the same, I had to test the bloom as it occurs naturally. To do this, we ran the following test:

1. The mage body pulled a mob. This put the mage on the mob’s threat list with 0.
2. I cast lifebloom on the mage. I generate 0.25 points of threat per amount healed with the HoT.
3. I let myself die, clearing my threat on the mob.
4. I resurrect and do merely one point of damage to the mob.

If the mob turns to me, the mage has less than one point of threat on the mob, meaning both the HoT and the end heal generated no threat for the mage.

As the video shows, I pull with merely one point of damage. This means the mage had 0 threat on the mob. This proves the “natural” end heal generates no threat for the recipient. However, it does not prove whether or not the natural bloom generates threat for the druid. For this reason, I turn to Xinhuan’s own testing:

Druid body pulls and takes some damage.
Druid casts Lifebloom on himself.
Druid clicks off Lifebloom after 1 tick. It healed for 67, no overheal.
Mage melees for 2-3 damage at a time.
Mage pulls aggro at 19 damage.

Conclusion: The Lifebloom HoT generates threat at 0.25 threat per 1 damage healed. We tested this with Rejuvenation as well, Rejuvenation generated threat at 0.5 threat per 1 damage healed. (67*0.25*1.1 = 18.425)

———————
Druid body pulls and takes some damage.
Druid casts Lifebloom on himself.
It healed for 67 for 7 times, followed by a 600ish final bloom, no overheal.
Mage melees for 3 damage at a time.
Mage pulls aggro at 130 damage.

Conclusion: The Lifebloom HoT generates threat at 0.25 threat per 1 damage healed to the druid. The bloom caused 0 threat to the druid. (67*7*0.25*1.1 = 128.975)

This reinforces my own tests of the threat generated by the HoT. Furthermore, it shows that the natural end heal generates no threat for the druid.

Special thanks to Xinhuan for inspiring more discussion and tests, as well as Althorn for helping me make the video.

Also, I have added this addendum to the original entry so people do not have to link to two entries. People need only to link the original article.

Back to Your Regularly Scheduled Blogging

Now that I’ve had my nitpick about this long-standing debate, I can go back to writing regularly. If there’s one thing I have to say: making videos is arduous and time consuming. Between two jobs, raid leading, blogging, and seeing The Dark Knight more than once, I simply don’t have time to be making videos for all of my entries. But it had to be done for this particularly sensitive topic.

Quite simply, I was sick and tired of people making their arguments haphazardly and erroneously, so I had to include something empirical that could not be disputed. Then again, even after posting hard evidence showing previous tests to either be flawed or erroneous, I still have people telling me I’m wrong.

That almost deserves an ASCII /palmface.

Almost.

In retrospect, we conducted this test sloppily. We were not careful to avoid threat anomalies and the argument had a couple holes in them. Regardless, I still believed our findings were rather conclusive, or at least more conclusive than those of others. However, some people on the EJ boards continued to argue that the end heal actually did generate threat for the recipient, because it displayed similarly in the combat log to prayer of mending—a spell that does generate threat for the recipient. This in and of itself is fallacious. Just because it appears in the combat log like a spell that does generate threat for the recipient doesn’t mean lifebloom’s end heal behaves similarly.

Flash forward to today, a year and four months later. Many patches and hotfixes have been implemented since the original test was conducted. Other tests have popped up in the discussion section of lifebloom on WoWWiki with the same conclusion as my own. But the debate rages on. Why? Because people see tanks pulling threat on the murlocs during Morogrim. What they fail to understand is that there are many different things going on during the fight that could potentially cause the warrior to pull aggro on the murlocs. Health stones, rage generation abilities and procs, PoM and earth shield procs, refreshing battle shout, etc. For this reason, people should not take their observations in such environments as suggestive of their beliefs.

Threat mechanics should be evaluated in a manner similar to how scientific tests are conducted. You wouldn’t study terminal velocity by pointing several industrial strength fans at a falling object, would you? I’d hope not. Neither would you test a chemical reaction between baking soda and vinegar by first mixing the vinegar with bleach (don’t do this; it creates a toxic gas). Similar logic should apply to testing game mechanics. You can’t accurately test the threat of specific components of lifebloom without simplifying the conditions and removing secondary and tertiary threat sources.

Being unhappy with my tests of old, and frustrated with the continuing debate, I have reconducted my test with results I feel are more accurate. Additionally, I have provided videos showing the premises used to reinforce my conclusion.

Initial Testing of the Threat on Lifebloom’s End Heal

Lifebloom is probably one of the more complicated spells in the game. Not only is there a HoT that can potentially cause threat for the druid, but the end heal has specific conditions for which it is triggered. In one condition, the HoT can run its full course and “blooms” into the end heal. However, the final tick of the HoT coincides with the bloom. So to simply put the spell on a warlock and have them life tap just before the final tick would create two simultaneous events (the HoT tick and the end heal) for which some value of threat is applied (even if it’s 0 threat, it’s still a value of threat).

Luckily, the end heal can be triggered by purging or dispelling lifebloom. And because there’s a one second delay on the first tick of the HoT, it can be purged or dispelled before it ticks. So the end heal can be isolated from the HoT portion of the spell by having an opposing shaman or priest on hand to dispel.

Unfortunately, however, there are a few actions we simply can’t avoid. These are as follows:

• The warlock must body pull a mob.
• The druid must taunt the mob off the warlock.
• The druid must apply lifebloom to the warlock.

Each of these events could potentially cause threat. However, thanks to extensive testing done over the years, it is known that body pulling only puts a person on the mob’s threat list with 0 threat. It is also known that taunting can put someone on the mob’s threat list and grants the taunting player with a threat value equal to the threat of the previously aggro’d target. So taunting a mob off someone who merely body pulled will result in having both players on the mob’s threat list each with 0 threat.

However, I advise being careful about anomalies that can occur performing these simple tasks. For example, if a protection warrior blocks after taunting, he or she could generate threat from a shield specialization proc. Likewise, having natural perfection proc could potentially generate threat. This video highlights this issue:

We also have to consider buffing someone with lifebloom itself. Does the mere application generate threat? In the second pull of the video above, you can see me casting lifebloom on myself. The mob does not switch to me until it ticks, suggesting the application generates no threat. But we need something more conclusive.

So what I ended up doing was this:

• A warlock body pulls a mob.
• The druid applies lifebloom to the warlock.
• The warlock removes lifebloom before it ticks.

Canceling lifebloom manually does not trigger the bloom. And because there’s a delay of one second before the first tick of the HoT, the buff can be clicked off before this occurs. So neither the end heal nor a tick of the HoT will happen, merely the application of the buff. So if the mob switches to the druid, the application of the buff generates more than 0 threat. If the mob does not, then the application of lifebloom generates no threat.

As you can see in the video, the mob does not turn to me after applying lifebloom, meaning it generates 0 threat. With this in mind, we can set up some conditions for testing merely the end heal of lifebloom. Here’s how the first test will be conducted:

• Have the warlock body pull the mob. This puts the lock on its threat list with a value of 0.
• Have the druid taunt the mob off the warlock. This puts the druid on the mob’s threat list with a value of 0. This also sets up a condition where the mob has the druid aggro’d, but the warlock is on its threat list, so any threat generating action performed by the warlock can be registered on the mob.
• Have the druid apply lifebloom to the lock while the shaman spams purge, causing the end heal to trigger on the warlock before the HoT ticks.

If the mob turns to the warlock, the end heal generates at least some amount of threat on the lock, because it would only take a value greater than 110% of the druid’s threat for the warlock to pull aggro. If the mob does not turn to the warlock, however, there are four possibilities:

1. The end heal generates threat greater than 0 for both the druid and the warlock, but equal for both.
2. The end heal generates threat for only the druid.
3. The end heal generates more threat for the druid than the warlock.
4. The end heal generates threat for neither the druid nor the lock.

Furthermore, after triggering the bloom, we will have the warlock generate a small amount of threat by casting fel armor. This ability generates 50 threat. So this would give us an idea of the difference in threat between the warlock and druid, even if lifebloom did generate threat for one or both parties.

As you can see, the mob does not turn to the warlock when the end heal is triggered. However, we also see that the warlock was able to pull the mob off the druid with merely fel armor. With this in mind, we consider each possibility from before:

1. The end heals generates the exact same amount of threat for both the druid and the warlock. If this is the case, then it generated between 1 and 499 threat for both, as 50 threat would cause the warlock to overcome the 110% threshold for any value between 0 and 499.
2. The end heal generates threat for only the druid. If this is the case, then it generated 45 threat or less for the druid.
3. The end heal generates more threat for the druid than the warlock. If this is the case, then, again, it generated 45 threat or less for the druid.
4. The end heal generates threat for neither the druid nor the lock. The warlock pulling the mob with 50 threat does not rule out this condition.

Coming to a Conclusion: Testing Possibilities #1, #2 and #3

In order to rule out the possibilities that: 1) the end heal generates the exact same amount of threat for both the druid and the warlock, 2) it generates threat for only the druid, or 3) the end heal generates more threat for the druid than the warlock, we must add a fourth person to the experiment: another druid. Why another druid? Because they are likely to have a weapon skill that will cause only one point of glancing damage to a high level mob. Here’s how the experiment will be conducted:

• A warrior will put on some extra gear, creating a health deficit so the end heal does not overheal.
• The warrior will body pull a mob, placing them on the threat list with 0.
• Druid A will taunt the mob, placing them on the threat list with 0.
• Druid B will also taunt the mob, placing them on the threat list with 0.
• Druid B will cast lifebloom on the warrior, and the shaman will spam purge on him.
• Druid A will then do exactly one damage to the mob.

If druid B has any threat greater than 0 on the mob, druid A will fail to pull aggro with just one point of white damage, as it is less than 110% of anything greater than 0. But if druid A does pull aggro, this means druid B has 0 threat. And because the warrior shouldn’t pull aggro from druid B (considering our earlier test), this would mean the recipient also receives 0 threat from the end heal. This will definitively rule out the three possibilities in question and leave only the possibility that it generates threat for neither the druid casting lifebloom nor the recipient. So, without further ado:

As you can see, I pull aggro with simply one point of damage. We can then make the following conclusion: The end heal on lifebloom generates threat for neither the druid casting lifebloom nor the recipient of lifebloom.

Special thanks for helping me conduct these tests are deserved by: Silverstreak, Fenaldor, Cuppycake, Aware and Shards.

An Addendum: Purged vs. Natural Blooms

The article assumes that the bloom portion of Lifebloom from being offensively dispelled is the same as the bloom that occurs naturally from one that naturally expires.
–Xinhuan in a comment on the original article.

This is a valid concern expressed about the original tests. So further testing had to be conducted to strengthen the argument and provide more empirical evidence suggesting the end heal generates absolutely no threat, regardless of how the “bloom” is triggered. So an addendum must be made. Such is the way of the scientific and dialectical methods.

To make certain the purged bloom is the same, I had to test the bloom as it occurs naturally. To do this, we ran the following test:

1. The mage body pulled a mob. This put the mage on the mob’s threat list with 0.
2. I cast lifebloom on the mage. I generate 0.25 points of threat per amount healed with the HoT.
3. I let myself die, clearing my threat on the mob.
4. I resurrect and do merely one point of damage to the mob.

If the mob turns to me, the mage has less than one point of threat on the mob, meaning both the HoT and the end heal generated no threat for the mage.

As the video shows, I pull with merely one point of damage. This means the mage had 0 threat on the mob. This proves the “natural” end heal generates no threat for the recipient. However, it does not prove whether or not the natural bloom generates threat for the druid. For this reason, I turn to Xinhuan’s own testing:

Druid body pulls and takes some damage.
Druid casts Lifebloom on himself.
Druid clicks off Lifebloom after 1 tick. It healed for 67, no overheal.
Mage melees for 2-3 damage at a time.
Mage pulls aggro at 19 damage.

Conclusion: The Lifebloom HoT generates threat at 0.25 threat per 1 damage healed. We tested this with Rejuvenation as well, Rejuvenation generated threat at 0.5 threat per 1 damage healed. (67*0.25*1.1 = 18.425)

———————
Druid body pulls and takes some damage.
Druid casts Lifebloom on himself.
It healed for 67 for 7 times, followed by a 600ish final bloom, no overheal.
Mage melees for 3 damage at a time.
Mage pulls aggro at 130 damage.

Conclusion: The Lifebloom HoT generates threat at 0.25 threat per 1 damage healed to the druid. The bloom caused 0 threat to the druid. (67*7*0.25*1.1 = 128.975)

This reinforces my own tests of the threat generated by the HoT. Furthermore, it shows that the natural end heal generates no threat for the druid.

Special thanks to Xinhuan for inspiring more discussion and tests, as well as Althorn for helping me make the video.

Disclaimer: This post may induce conceptual and mathematical headaches.

Threat

“Threat” is a numerical value representative of the amount of hate a person has on a mob. After extensive evaluation of WoW’s threat mechanics, many players interested in the system have agreed to normalize threat around one point of “white damage” done to a mob. This means meleeing a mob for 100, without any threat modifying stances or talents, would generate 100 threat for a person on that mob.

Mobs can also use abilities that directly affect threat. For example, a mob can cast a spell that lowers a person’s established threat by a fixed percentage.

Threat Modifiers

Some classes have abilities and talents that affect their threat by a specific percentage. These percentages can create what are known as “threat modifiers.” For example, druids have a talent called subtlety that reduces the threat their spells generate by 20% at its maximum rank, meaning a modifier of 0.8 is applied to the threat of all the druid’s spells.

Many abilities have an innate threat modifier based on their threat production relative to unmodified white damage. For example, one point of effective healing done by healing touch generates threat with a 0.5 modifier. So if you heal a person for 5000, none of which is overheal, you will generate 2500 threat on a single mob. Holy light, on the other hand, has a modifier of 0.25, so a paladin who effectively heals someone for 5000 with holy light would only generate 1250 threat.

Furthermore, some abilities can have different modifiers for each component of the ability. For example, siphon life generates threat for its damaging component, but its healing component generates no threat.

Mobs can also use abilities that apply threat modifiers to people. For example, Kalecgos can place a debuff on someone that doubles their threat generation for a fixed amount of time.

Threat Lists

Each mob has what is called a “threat list.” A threat list is a catalog of players for whom the mob will react to threat generation. Certain actions can put a person on a mob’s threat list, including:

• Body pulling the mob.
• Damaging the mob.
• Putting a debuff on the mob.
• Buffing someone already on the mob’s threat list.
• Healing someone already on the mob’s threat list.
• Taunting the mob.

This list can vary, depending on how each mob is coded. For example, a mob could potentially be designed to use an ability that affects anyone in a given radius, and this mob could put those affected by the ability on its threat list. However, such abilities do not always put a person on a mob’s threat list, even though it might put them in combat. Many world mobs with demoralizing shout put anyone affected by it in combat without putting those people on their threat lists, for example.

AoE Threat

It is important to note some abilities divide their threat generation based on the number of threat lists on which a person resides. For example, a druid on the threat lists of ten mobs will have their threat caused by healing touch divided by ten, spread evenly across each mob. So effectively healing someone with healing touch for 5000 will generate 250 threat for each mob that has the druid on its threat list.

Healing, power gains (energy, mana and rage) and AoE buffs like battle shout typically divide the threat generation by the number of threat lists. But this might not always be the case, given WoW’s mutability and unique cases. For example, while the mana gained from the proc on Insightful Earthstorm Diamond generates threat at this point in time, the mana gained from life tap currently generates no threat.

AoE damage is typically not divided by the number of targets affected. This was not the case with some spells in the distant past, but most have changed to reflect this fundamental concept.

Aggro

A person with “aggro” is a mob’s primary target. This is not always the person highest on a mob’s threat list, however. This is because there are different values typically required to pull aggro from someone else. For a person in melee range of a mob, they must usually exceed 110% of the threat attributed to the player currently with aggro. And for a person outside of melee range of the mob, they must typically surpass 130%.

It’s important to note that a person with 120% of the currently aggro’d target’s threat will pull if they duck into the mob’s melee range, even if they do nothing before and after crossing into melee range.

Some mobs use abilities that can drop the standard requirements to 100% or below. Knockbacks used on the current player with aggro will typically do this. So even a person below the traditional ranged and melee thresholds, they can still pull if they are above 100% of the aggro’d target when he or she gets knocked back.

It is important to note that anyone simply targeted by a mob does not necessarily have aggro on that mob. Mobs will sometimes target a person to cast a spell on them, but they subsequently return to the player with established aggro once the cast has complete. Likewise, spells that cause the mob to select no one can work similarly and it is not always an indicator of someone losing aggro on that mob.

To further complicate the idea of aggro, some mobs do not actually aggro anyone nor follow the traditional guidelines of threat. Some mobs indiscriminately pick targets to cast spells on and cannot be forced to attack anyone highest on threat. Most prominent of these mobs in the current expansion is probably Dorothee and the Shade of Aran in Karazhan.

Taunting

Warriors have an ability called “taunt.” This ability forces a mob to aggro the warrior for a given amount of time. It also grants the warrior threat equal to the value of the previously aggro’d target before the taunt. If the warrior already has more threat than the person off which they taunted, however, their threat does not change. Paladins and druids have righteous defense and growl respectively. These abilities are simply variations of the warrior taunt and apply the same concepts of threat.

A person cannot pull aggro from a person taunting as long as the taunt debuff remains on the mob. However, once the taunt debuff has dropped, the traditional aggro thresholds of 110 and 130% apply.

Threat Anomalies

A “threat anomaly” occurs when threat behaves or mobs react to threat generation in an unexpected manner.

There are many game mechanics that can have strange effects on threat. For example, some adds during boss encounters will often spawn and ignore all threat generation mechanics for a short period of time. This is prominently evident during the Shade of Akama encounter in Black Temple, as the door adds spawn and ignore any threat generation for a short period of time. On the other side of the spectrum, some adds during encounters spawn with an established amount of threat on random targets. This is most evident during Solarian in Tempest Keep (or at least was when I was doing the encounter over a year ago).

Such examples no longer produce “anomalies,” as they are currently understood mechanics. Rather, variables that are still unknown cause anomalies. As new patches are introduced with new proccing gear, new proccing meta gems, new proccing enchants, and changes to abilities, new anomalies are always introduced and can take a lot of time to dissect between each patch. And sometimes various anomalies are debated exhaustively without any real conclusion. I believe how threat affects Patchwerk and Supremus‘ hateful strikes is still argued to this day, because it is also affected by people’s current hit points, and possibly people’s positioning and how latency can affect the server’s registration of each player’s position.

Testing Threat

The only way to accurately test threat mechanics is to do so by first understanding conditions that create anomalies, or by removing the conditions that create them. Unfortunately, this is difficult given certain environment. Getting 25 people to risk wiping because you want to experiment with Supremus’ hateful strike is near impossible. And you can’t exactly remove the positioning and latency issues unless you bring enough healers to keep the tanks and melee alive when the fire spawns right under their feet. So people often use Patchwerk as a reference point, but there’s a possibility his and Supremus’ hateful strikes aren’t similar enough to make perfect parallels.

One of the reasons threat mods like KTM and Omen have never been 100% accurate is because recognizing, understanding and removing anomalies in every case is an endless feat. Furthermore, coding bugs can create erratic behavior in various running scripts and change the results of a given test. But you can at least increase the accuracy by reducing anomoly-creating conditions and conducting repeated tests. For example, you can trigger the end heal on lifebloom and prevent the HoT from ticking by purging it with a shaman of the opposing faction, allowing you to test the end heal’s threat directly without the HoT creating secondary threat during your test of the end heal. This type of testing is important when dissecting the threat of a given spell or ability.

Conclusion

“Established” testing might not always be accurate. Unless repeated and extensive experiments have been conducted and reported with no logical fallacies, you shouldn’t always take what is written as gospel. “We tested the threat of the end heal on lifebloom and concluded that it does not generate threat” is not evident of fact. It does not provide empirical evidence. It does not detail the experiments used to prove the claimed law. It is simply conjecture.

Casting doubt on the results of someone’s test is part of the scientific method. It is an important step in the refutation or reinforcement of particular concepts and conclusions. The empirical evidence taken from an experiment is the foundation on which an argument should be built. And if the foundation itself is flawed, then so too is the argument. But if the foundation is sound, the argument structured well, and all doubt has been exhausted, then it can be written into law.

Until Blizzard patches the game and changes the law itself.

Healing Meters at a Glance

Before you begin to dissect this parse, you should know there are actually six healers and two shadow priests (Shards and Lightofwest).

At first glance, it definitely seems like we either have awesome druids or terrible shamans. If I went to the EJ forums and asked what’s wrong with this parse, a lot of people would respond, “Your resto shamans should be outhealing your druids.” But, aha! That’s the wrong assumption to make.

First and foremost, it should be noted this parse comes from M’uru. In this fight, you have four different tanks in phase one: one for the void sentinels, one for the void spawns (who appear in the location where each void sentinel dies), and one for each set of “door adds” (there are two sides, each including a caster and two melee mobs). Some people do this differently, but that is how we do it. Typically, as well, someone other than the traditional tank might soak damage from the caster mob on each door. I would elaborate, but I’m cautious of my server’s competition stealing our exact strategy.

Because there are so many tanks, our healing strategy is such that our druids are rolling four different stacks of lifebloom on people taking near constant damage. Considering our gear, both mine (Lilume’s) and Siafu’s triple-stacked lifebloom ticks for over 900 each. Mine typically ticks for about 950 (per second). So four of my triple-stacked lifeblooms produce a total of 3800 potential healing per second (HPS). Of course, however, the actual healing is much less, considering there are periods where the tanks are topped and a lifebloom tick heals nothing. Furthermore, there is a second phase where there is only one tank, so we switch to spamming regrowth on the raid, which has a lower HPS potential in general, but is more reliable in terms of saving people who drop dangerously low in that phase. So, across both phases of the fight, my HPS averages out to about 1653. That’s certainly less than what a resto shaman can produce on certain fights like Teron Gorefiend, but it is also important to point out that the raid-wide damage on this fight is not as heavy as the raid-wide damage you see during Teron. So a resto shaman’s HPS is going to be somewhat lower than it would be on a few other fights.

Next, you’ll note that our shadow priests outhealed our holy/discipline priest. That’s not supposed to happen, right? Well, both of our shadow priests were running improved vampiric embrace (VE). And in any fight where the entire raid is taking a constant stream of damage, and the holy/discipline priest is assigned to a tank, you’re going to see shadow priests with improved VE outheal them if the VE can outpace the damage the tank is taking. Also, it would be irresponsible of the priest to try to pad the meters by healing other people, considering a lack of his attention can cause his tank to die.

In this context, it would be erroneous to make the assumption that our druids are awesome and our resto shamans are terrible, and a mistake to assume our holy/disc priest sucks. In truth, healing meters merely reflect the trends set by both the healing strategies and the features of each encounter. This is the one thing most raid leaders and recruitment officers fail to recognize and I cringe every time someone tells me “healing meters matter.” Yes, they matter, but the point at which they matter is very subjective. It’s very important to consider the situational strengths of each healer, the task(s) they’re given, and what tools they use to heal.

Mana Potions at a Glance

This screenshot shows the number of times someone used a mana potion. Now, considering Siafu and I performed similar roles, it can be argued that Siafu should have used three potions instead of two. Considering the fight lasts over seven minutes, he should have had more than enough time to chug another mana potion. We do spam regrowth for a solid 90 seconds in phase two, afterall, and using only two mana potions puts him at risk of running out of mana before the fight is done. So the optimization of using mana potions is extremely important, and he probably should have planned to have his third mana potion available either for the start or the middle of phase two.

However, it’s also important to consider what exactly each person in the raid was given to work with. For example, if one of the resto shamans had a shadow priest (and one of them did), it is likely he would have only had to chug one or two mana potions to remain at full mana going into phase two. This could have potentially provided him the opportunity to chug a haste potion for the final 15 seconds of the fight instead of a mana potion.

Miscellaneous Information (Using Innervate as a Specific Example)

It would take me pages upon pages to cover every single aspect a WWS parse can tell you about what your healers are doing right and wrong. You have to look at mana potions, water shield, mana tide, mana stream, shadowfiend, symbol of hope, trinkets, vampiric touch, gear choices, flask and elixir choices, gem choices, weapon oils, talents, innervate, etc. But, if you’re considering deaths, you also have to look at things outside of your healers’ control: healthstones, last stand, shield wall, shield block, shadow word: death, recklessness, deathwish, ironshield potions, etc. And this doesn’t include encounter-specific issues like people not standing in cave-ins during Gruul.

So, for the sake of simplicity, let’s look at the use of innervate from the same M’uru parse used earlier.

Only one innervate, which came from myself. At this point, the alarm bells should be ringing for a raid leader. There’s five druids in the raid (two resto, one balance, and two feral), and the feral druids weren’t tanking for phase two. But only one innervate was tossed out. There should have been at least five. Now, this was something we talked about after the attempt. And we did manage to rectify the problem the next time we saw phase two.

Much better, but not perfect. It should be noted that this attempt on phase two was shorter than intended. We made some errors on our positioning of the boss and this wiped us. So we never got a chance to use our fourth and fifth innervates. There was also some confusion because I had forgotten which person I assigned myself to innervate. So I innervated Dess, instead of Lightofwest. And the person who was supposed to innervate Dess was confused as a result, especially when someone else innervated Lightofwest. I also forgot about our balance druid and didn’t assign him an innervate. I probably should have assigned him to Caiyn, in retrospect.

This is an example of how a single element can add complexity to how people perform in a raid, however. If you plan your innervates and someone expects to receive one, they often shift their approach to the encounter with this in mind. Healers will overheal more often knowing conservation isn’t as big of an issue when they will be the target of an innervate. Without one, however, the healer will run out of mana sooner than expected or will adopt a mindset rooted in conservation. This can then put the tank he is healing in considerable danger, with the healer either running out of mana or downranking to conserve.

Now, consider that innervate is only one element in an entire raid. Factor in all of the other elements, and suddenly healing data becomes so complex you can barely wrap your head around it. For this reason, I value specific observation of performance and evaluation of all the elements in a raid over simply looking at the meters, cooldown use, etc. If the tank someone was assigned to died, but the tank missed a spell reflection on a mob that hits with a 10K damage spell, I shouldn’t hold the healer responsible as a raid leader. I should first question what happened leading up to that death, and rely on things like player feedback and GrimReaper to provide information about what might possibly have gone wrong.

Basic Tips for Assessing Healers

1. Specifically watch a healer’s performance in conjunction with the performance of the people they’re assigned to.

Don’t just rely on healing meters and deaths to asses performance. While their placement on healing meters can be correlative to good or poor performance, this is not always the case, due to reasons stated in the earlier sections of this entry.

Check to make sure they’re adhering to their assignment and understand why they’ve been given that assignment. Make sure they’re properly timing and using their potions and cooldowns. Make sure they’re in a group that compliments their role if needed (if it can be afforded). Make sure they’re downranking and upranking responsibly. And make sure the players they’re healing are doing what they need to do to make the healer successful at his or her job. You can’t tell if a healer is playing well or not if your tank dies because he can’t find his shield block, shield wall, last stand, nightmare seed, spell reflection, or healthstone hotkeys when needed.

After this, watch them. If you see them standing there as their assigned target slowly loses health, there’s something wrong.

2. Make sure they provide feedback relevant to the healing strategy.

Healers who actively contribute feedback relevant to the healing strategy are an extreme asset. If they are finding it difficult to keep a tank up because they have to downrank to avoid running out of mana, make sure they tell your healing strategist. This can be important to making group adjustments that compliment the need to uprank further. Or, if what they’re doing is keeping the tank alive with no mana problems, make sure they tell you this as well. You can potentially free up space in group that would be more synergetic with somebody else therein.

3. Make sure they’re using the right gear for the fight.

If they’re complaining about mana, and HPS isn’t the issue, and they’ve stacked too much haste, figure out why. Maybe they don’t have the gear to swap in and out to shift their style more in the favor of longevity. Or maybe they had the opportunity to and they made poor gearing choices by letting stuff rot, or poor gemming choices.

The progression of your guild can make or break applicants in this regard. I know I wouldn’t want a healer who is wearing nothing but haste gear in Sunwell. It certainly has its place, but screw him running out of mana on fights grounded more on longevity than HPS potential.

4. Sometimes healing assignments are actually the problem.

There are fights in Sunwell where healing assignments have to be so finely tuned that anything your healers do won’t possibly be enough if you don’t micro-manage enough. So sometimes failure is at the hands of the assignments, rather than healer performance. That said, however, it is in your best interests to possibly recruit a good amount of healers or players in general that can help point out any oversights if you make a mistake.

5. Sometimes raid or group composition is the problem.

No druids on a high-mobility fight? No resto shamans or circle of healing priests on heavy raid-damage fights where players are in close proximity? These are two examples of raid composition problems. Your healers likely won’t be at fault for people wiping in this case.

Conclusion

Assessing your raid’s healers is never as simple as just studying the meters, cooldowns, downranking, and talents of your healers. Raid and group composition, healing assignments and strategies, and even the performance of non-healers play a part in your raid’s general ability to stay alive. If people can’t get out of cave-ins on Gruul during high growths, they’re going to die, regardless of how good your healers are.

When someone places low on the healing meters, it might not always be their fault. A priest spamming circle of healing is typically going to run out of mana eventually if they don’t receive a shadow priest or an innervate. But meters can sometimes indicate performance, so don’t ignore them entirely. However, you have to assess the entirety of the situation before you can understand whether or not the healer is at fault or if there’s a flaw somewhere else.

Assessing and directing your healers is a complex job, and not everyone is cut out for it. You really need a strategic mind to fully understand it and no amount of talking by me will fully teach you how to lead your raid’s healers and make judgments on their performance, especially because strategies change per encounter. On one fight, you might have ten healers, while on another you might have six because you need more DPSers to make the DPS budget.

It’s not an easy job, but these tips should help you get started.