Friday, September 28, 2007

Seeking Divination in Retribution Changes

So far, Blizzard has announced the following changes for the paladin class:
  • Mana cost for Exorcism, Holy Wrath, and Hammer of Wrath lowered
  • Vengeance duration increased to 30 seconds
  • Improved Seal of Crusader benefits put into base spell, talent benefits replaced with those of Sancitified Crusader
  • Sanctified Crusader renamed to Sanctified Seals, which increases chance to critically hit with all spells and melee attacks by 1/2/3% and reduces the chance your Seals will be dispelled by 33/66/100%
  • Crusader Strike cooldown reduced to 6 seconds
  • Vindication frequency and duration increased and reduces all attributes by 5/10/15%
  • Pursuit of Justice is now 3 ranks and increases movement speed by 5/10/15% and reduces the chance to be hit by spells by 1/2/3%.

These changes are interesting. First, the change to Vengeance means that a lower crit percentage will be required to maintain perma-Vengeance, and that Vengeance can be maintained even on fights which require a little running.

The Vindication, Crusader Strike, Pursuit of Justice, and Sanctified Seals changes are significant buffs to PvP Retribution paladins.

The change to Improved Seal of the Crusader is interesting. It takes the one major raid buff that Ret paladins have and puts it in reach for Holy or Prot paladins to pick up.

However, the glaring omission is threat reduction. Right now, PvE Retribution paladin DPS is limited more by threat than anything else. Without a threat reduction mechanic of some sort, these changes will not increase Ret paladin DPS.

However, these changes do open the door to a potentially more interesting possibility.

Disclaimer: This is complete speculation, and probably isn't correct.

Lets look at the breakdown of Retribution DPS. Assume that the paladin does 100 points of damage and is at the threat limit. From my experiences, the damage breaks down as follows:

- 55 points from Auto-attack
- 18 points from Crusader Strike
- 18 points from Seal of Command
- 9 points from Judgement of Command

Now, the change to Crusader Strike means that Crusader Strike damage will jump from 18 points to 30 points. But the paladin is still threat-capped at 100 points. So the new breakdown will look like:

- 55 points from Auto-attack
- 27 points from Crusader Strike
- 18 points from Seal of Command

What does this mean? It meanst that the paladin no longer needs to Judge Command.

This has two important ramifications:

1. You can use Seal of Command (Rank 1). Combined with no judging, This means Retribution DPS becomes very mana-efficient when threat-capped.

2. 100% of your damage scales with Attack Power. Retribution does not need Spell Damage anymore.

So what should Blizzard do with all that spell damage on Retribution? You could add extra Strength, but you're already threat-capped. Instead, you could replace it with straight +healing. Retribution gear would feature Str, Sta, Int, and +healing.(The fact that +healing now gives a little boost in spell damage only makes this option better.)

So now we have a mana-efficient melee character, with a decent amount of +healing, and who is expected to have to hold back DPS because of threat issues. Add to that the fact that Vengeance now lasts twice as long. This allows the paladin to break off DPS and toss a heal without Vengeance falling off, while staying near the threat cap.

Perhaps this is Blizzard's latest attempt at creating a paladin who melees and heals.

Crazy? Maybe. Reality? Probably not. Madness? Sparta. Worth trying? Definitely.

Wednesday, September 26, 2007

Retribution using Seal of the Crusader?

A reader writes in:
I have heard a lot of outcry for some sort of heal-per-dmg component to the ret paladin, as some people see that all we bring is an extra 3% crit to the raid (which in my opinion is invaluable depending on your raid group). I however am in the process of purchasing my Khorium Champion, and then the mats for a Battlemaster enchant.

I also plan on using a modified physical damage build with seal of the crusader ALWAYS on to get as many procs from the weapon, the enchant, and of course the wisdom and light that I can keep indefinitely judged on mob from any other paladins in the group. the build I have right now is linked here. I was wondering what you thought about it. I have aura mastery to benefit caster dps who stay at their max ranges from Sanctity Aura. I was wondering where you think those last few talent points should be allotted to.

I realize the nerf to my possible damage output by using a less than 3.6 speed weapon and not using Seal of Command, but I feel like those are tradeoffs I would gladly take to see the amounts of stacking heals from procs, and judged light, and mana recieved from judged wisdom.

I don't really think that this is a good idea. The problem is that you give up a *lot* of damage by not using Seal of Command. As well, SotC does not improve as you get better gear, while SoC does, so as time goes by you are giving up more and more damage.

The trade-off is something like losing 30% of your damage for extra Light, Wisdom, and Battlemaster procs. If it was a smaller number, like say 5% or less, then it might be worth experimenting with. But something like 30% is way too high to lose.

If you take a DPS slot in a raid, your first priority is to provide massive amounts of damage. Paladins already do lower DPS than other classes, and what you propose would lower that still further.

What this type of playstyle is really good for is grinding or farming. Here you can use the extra procs from Wisdom/Light to keep your mana/health up and reduce downtime.

If you have a slow green or blue 2H, try using both these strategies on some mobs with a damage meter enabled. (People say that you can use the Deathstalker in Shadowfang Keep, I've never tried this.) SotC will add the same amount of damage to a slow weapon as it will to a fast weapon. So you can see how much DPS you do with SotC vs SoC, and you can see if the trade-off is worthwhile. Just keep in mind that the gap between the two will keep growing as you get better gear.

Then you can make a decision on whether to invest in the sword and the enchant, or if there are better weapons out there.

Utility is good, but in the end you will live or die by the DPS you put out. For a DPS player--which is what a Ret paladin aspires to be--maximizing DPS is more important than maximizing utility.

Those are my thoughts, but my advice is to try them both out (before buying the expensive weapon!). It could be that I am wrong, and the difference between the two is much closer than I think it is.

Tuesday, September 25, 2007

Patch 2.2 First Impressions

Ahh, patch day raiding. So much troubles. My raid UI was completely messed up with groups being stuck in the middle of the screen. If you have that problem, type /reloadui until you can click on them, then close them immediately. After doing that several times, things got cleaned up nicely.

Also, my UI scale seems off. Everything seems very slightly larger, and things are not fitting together quite perfectly.

I think the new sound engine is working well. Everything seems to sound a little bit better to me, and the game performed slightly better. As well, there are new sounds scattered about. I haven't tried the Voice Chat yet.

The buff changes are extremely nice. 10 or 30 minute buffs are great. Also, the changes to pet buffing are really nice. If you're wondering how pet buffing with greater blessings works, it goes like this:

1. Hunter pets share blessings with Warriors. Buffing a warrior will hit all the warriors and hunter pets. The jokes just write themselves.

2. Warlock pets share blessings with Warlocks. Buffing a warlock will hit all the warlocks and their pets. The only exception is a phase-shifted imp. If the imp is phase-shifted, it will not receive the buff.

This is a great change, as it means that the pets get buffed without any extra effort on the part of the paladin.

The nerf to Blessing of Sacrifice was muted. The cooldown was reduced to 30s, which is the duration of the spell. So that means you can still use it in PvE on tanks. It is still useful in PvP, but it can be countered by Purge.

I haven't tried out the new Seal of Vengeance yet, so I can't comment on it.

Other than that, the patch looks pretty decent. Many people have commented that it's not worthy of being a "full" patch, as it doesn't have a lot of new content. However, I think introducing a new sound engine is a big enough change that it warrants its own patch.

Monday, September 24, 2007

Raids and Trash Respawns

Questionc, in a comment to a previous post, writes:
The way you describe the trash spawn, it doesn't really sound any different from respawns used in any other part of the game... as an instance time limit. A good example of this is Scarlet Monastery, where failure is taking so long that you have to fight to get out of the instance.

And honestly, it's a good gentle mechanic. It's fair for them to put a time limit, and much nicer than just kicking you out of the dungeon if you take too long.

The problem with this is that there is a fundamental difference in expectations between raid instances and 5-man instances.

When you go into a 5-man instance, you expect to clear it, even if it is your first time going to that instance. You may wipe once or twice, but the general expectation is that you will finish that instance. In this case, an instance time limit is a good idea. If you hit the trash respawn, it's a sign that something is wrong, and perhaps it might be a better idea to try again later, maybe with a different group or after gearing up a bit more. If you wipe enough that trash respawns, it is unlikely that your group will successfully complete the instance.

In contrast, in a raid boss fight, you expect to wipe a lot. That the fight is intricate enough that learning and mastering it will take multiple hours. It took us two or three evenings to refine our strategy and master The Lurker Below, and he's generally considered an easy boss in SSC.

Respawns, rather than being a time limit, are interruptions in the learning process. As well, they are interruptions which eat up a lot of valuable time. If you raid from 6-9pm, and you get respawns at 8:30, you may as well call the raid. That means you lost half an hour of learning time, which is a lot for a casual guild.

Respawns are annoying because they make you waste time. Time is the one thing that casuals do not have a lot of. It takes a lot of coordination and administration to gather the people necessary for a raid, and some of that work gets tossed away because Blizzard decided to have respawns.

If we didn't have respawns, we could make better use of the time we have available, we could accomplish more. Probably not a whole lot more, but still. Trash respawns are obvious busy-work. We've already proven that we can defeat them once, why do we have to go through it again? We want to focus on the boss, why not let us focus on the boss?

The point is that respawns are unnecessary. Different guilds have different time constraints. Given that learning bosses is an activity which requires multiple hours, it would be best if guilds could raid to match their own constraints, rather than having to deal with frustrating, artificial constraints as well.

As for the idea that respawns are needed to slow guilds down, I don't think that is true. Respawns don't stop the elite guilds. Multiple guilds have already completed the raid content in WoW. Why slow down the non-elite guilds to an extra degree? Solid and demanding boss fights will keep a damper on progress.

I guess it boils down to the fact that I really don't like "artificial" barriers to content. The content itself should provide the challenge. The challenge in raiding is mastering the boss fights, not clearing the respawned trash that you've already beaten once that evening.

(Which is not to say that there should be no trash. Trash provides little "mini-challenges" and makes the instance feel alive. But once the challenge has been met, that's the end of it, and the trash should stay dead.)

Sunday, September 23, 2007

Priests are the Hybrids of Healing

I had an amusing thought about healing Priests in WoW.

Traditionally, the problem with hybrids is that while they can do multiple things, they can't do them as well as a pure class. You can tank, heal, or DPS, but not as good as a pure tank, a pure healer, or a pure DPS. And when putting together a raid, rather than taking two hybrids, you get better performance by taking 2 different pure classes.

If you look at the healing classes, you can see the same pattern emerging. Paladins are the best single-target healers in the game. Druids are the masters of heal-over-time spells. With Chain Heal, Shamans are number one when it comes to healing splash damage.

Priests can do all these things. They can direct heal, they have a solid HoT, they have decent splash heal tools. But they aren't the best in any one category. They're second-best in all 3.

Priests are the hybrids of healing. And thus they suffer from the traditional problems of a hybrid class.

Thursday, September 20, 2007

The Lurker Below

We killed The Lurker Below in Serpentshrine Cavern last night. It's an interesting fight. Basically, Lurker is a giant fish and you fight him on a couple of islands. He does a "spout" which is basically a beam of death. You have to dive underwater when he spouts to avoid it. If you kill all the trash, the water around Lurker is superheated, and does 500 damage every couple of seconds. If you don't kill the trash, the water has these crazy fish which will kill you very fast. There's also a phase with naga adds. It's a control fight more than a burn fight.

I really hate trash respawns though. The trash respawned after a few attempts, while we were still on Lurker's island. So we did this very complicated warlock summoning chain to get back to the platforms where the trash were and killed them again before finally downing Lurker.

Seriously, Blizzard, trash respawns are a terrible mechanic. Just let us work on the boss in peace and at our own pace. Heck, because of respawns we stretched the raid out an extra hour, just because we were so close. If the trash hadn't respawned, we could have killed Lurker and ended the raid on time.

Most guilds have their time constrained by outside, non-game factors. Adding extra time constraints in-game is unnecessary. The elite guilds who don't have outside time restrictions don't really care about trash respawns, and it's been shown to have very little impact on them. So why punish the non-elite guilds like this?

Monday, September 17, 2007

Theorycraft vs Trial-and-Error

In response to the previous post, Jason comments:
Hmm what this is showing me is that DPS actually takes a lot more brains and initiative to play well. More than it does to play a tank or healer well. As tank and healer might have results slapped in front of them, so they can correct easily.

The DPS game is more subtle and requires keener observations and more diligent research.

This is not quite true.

(As an aside, I don't really understand why people care if DPS or healing is innately harder or easier. I care if one is unnecessarily harder.)

There are generally 3 ways that people learn to play:

1. Theorycrafting. This means looking at the math behind the game, and extrapolating tactics based on the math.

2. Trial and error. The player tries something new, sees if it works better than the old way, and decides whether to use it or not.

3. Appeal to authority. Get someone who is better to tell you what to do.

The third one inevitably chains back to the first two, so we'll discount it for now.

Of the first two, most people learn through trial and error. As they get new abilities, they try them out, and based on their experiences they use the ability or not.

A few people, though, are theorycrafters. They build mathematical models, and the models imply that certain abilities should be better than others. They test out the abilities and use the results to improve the model.

Certain activities are easier to learn or improve with different strategies. For example, crowd control is generally learned through trial and error. The difference between a good trapper and a poor one is more a matter of practice than theorycraft. Healing is very similar. Theorycraft helps, but the better healers tend to be the ones who heal a lot.

Trial and error requires feedback. Test something, see the result, incorporate the change. If you can't see the result, or if the result is hard to understand, you can't use trial and error to learn something.

So raid DPS is hard to improve if you use trial and error, like most people do. However, doing good raid DPS is relatively simple from a theorycraft point of view. Where the theorycrafters go into their long discussions and crazy math is determining the optimal raid DPS. But you don't need optimal DPS to start with, you just need good DPS.

Here's a general algorithm for doing high DPS:

1. Rank your spells from highest DPS to lowest DPS.
2. Cast the highest DPS spell.
3. If you cannot cast the highest DPS spell again (because of a cooldown), cast the next highest.
4. Repeat steps 2 and 3, going down the list as necessary.

It seems very simple, but following this will provide you with a decent spell rotation for every class. It won't be the optimum rotation, and you might have to drop a spell because it costs too much mana, or does too much threat, or has a really short range. The work comes in ranking your spells correctly. (The main trick here is that when calculating DPS for a non-burst fight, you only look at the cast-time of the spell, not the lifetime of the spell. For direct damage they're the same thing, but not for DoTs.)

This is a good starting algorithm to develop a rotation. After this, you have to consider "enablers" such as Mangle, or Improved Scorch. These are spells which have a lower DPS, but improve spells above them on the DPS chart. Then you have to see if the DPS lost while casting the enabler is made up by the gain on the rest.

And then after that, you have to consider mana costs, which is where it gets complex. But getting a decent DPS rotation is not very hard from a theorycraft point of view. (Getting the single best rotation for a given fight is where the fun lies.)

A very simple example is the Frost mage rotation. If you ranked the spells, it would probably look something like:

1. Summon Water Elemental (3 min/45s cooldown)
2. Ice Lance (against a frozen target)
3. Frostbolt
4. Fireball
5. Ice Lance (against a non-frozen target)

So when looking at the DPS rotation, you go Summon Water Elemental first. Then you can't cast Summon Water Elemental again, so you move down the chart. The boss isn't frozen, so you move down to Frostbolt. You can cast Frostbolt again and again, so you don't go further down the chart. When Summon Water Elemental comes off cool-down, you cast it again.

That's a relatively simple rotation (in the game you should also Cold Snap to reduce the cooldown on the elemental). But the basic principles apply to every class.

So that was a really long post, but it essentially boils down to: raid DPS is easier from a theorycraft point of view; healing is easier from a trial and error point of view; and tanking is a bit of a mixture of the two. The problem is that the majority of people learn using trial and error, not theorycraft.

Saturday, September 15, 2007

A Perfect Example of Improving DPS

BigRedKitty posted a perfect example of what I was talking about in the last two posts.

A hunter named Jeor has been trying to improve his DPS. His complaint:
So again I have to ask, what am I doing wrong? what is the secret that no one is telling me? is it my shot rotation? something about my build or gear? is there a special macro that no one is telling me about?

The key element here is that Jeor doesn't know what he's doing wrong. He's not stupid or unwilling to improve. However, the game has not provided him with any of the feedback he needs in order to improve.

BRK talks to him, tightens up his spec (Beastmaster), changes his shot rotation, and does a little re-gemming, and the results are staggering. Jeor provided WWS parses for Gruul attempts for the week before his tweaking and the week after. He jumps from 451 DPS to 964 DPS, a gain of over 500 DPS!

Look at Jeor's first week shot mix. Not using Steady Shot enough? Check. Missing shots? Check. Not using Kill Command? Check. The same hunter mistakes, again and again.

Compare to his shot mix the next week. Now that's acceptable DPS.

BRK's post was very useful precisely because of the WWS parses, only 1 week apart. Feedback improves your play, and DPS players don't get feedback from the game, so they have to go outside it (to BRK in this case) in order to get the feedback they need to improve.

Wednesday, September 12, 2007

Response to Comments on Bad DPS

(See the previous post.)

First off, believing that "most people are dumb" is an adolescent fantasy. Most people are average and want to get better. They just don't know how. What works in one area doesn't work in another area, and that dissonance causes them problems.

If you're a guild working on T4 content, take a WoW Web Stats log of your raid. Or take a look at one of the numerous WWS posts on the Raid and Dungeons forums. Over and over you will see the same mistakes. And it's not even "mistakes" like frost vs fire.

You'll see:
- Hunters not using Steady Shot
- Hunters missing shots
- Rogues using Shiv instead of Sinister Strike/Backstab
- Dagger rogues using Sinister Strike
- Rogues missing attacks
- Rogues using Eviscerate instead of Slice and Dice/Rupture
- Warlocks not keeping DoTs up
- Cat Druids not using Shred
- Mages missing spells
- Mages using many different spells instead of their best spells
- Shadow Priests not keeping DoTs up

You'll see the same mistakes, again and again. And at some point, you have to wonder if there's a reason that all these different people are making the same mistake, and how things could be changed to keep them from making such systemic mistakes.

Also, I'm not saying that healers and tanks are perfect. It's just that our mistakes tend to have immediate consequences, and thus we try to correct them on the very next attempt.

You do occasionally see systemic tank and healer mistakes. Tree Druids who don't stack Lifeblooms, Priests who spam Flash Heal, Prot Warriors who don't Shield Slam.

(The fact that Paladins only have 2 healing spells, one of which is spammed on a tank, makes it pretty hard to have systemic errors. About the only one I can think of is paladins not using Lay On Hands.)

Healer/tank mistakes tend to be of the immediate, tactical variety, such as healing the wrong person, or BoPing a warlock with Moroe's Garrote instead of saving it for a healer/mage. The thing about tactical mistakes is that you realize the mistake quite soon after you make it. That "feedback loop" is there, and that makes it easier to improve.

Tuesday, September 11, 2007

Why are DPS Players so Bad?

One of the largest problems in early TBC raiding is DPS players. Quite frankly, most DPS players don't seem to know how to do acceptable amounts of damage.

This wasn't an issue in pre-TBC raiding because the early WoW 1.0 raid DPS requirements were abysmally low. It wasn't until you got to AQ40 that DPS had to start performing anywhere near their full potential.

Let's take Gruul the Dragonkiller. If your DPS does an average of 500 DPS, you are looking at a growth 16 kill. If your DPS does an average of 600 DPS, you're looking at a growth 12 kill. Is asking for 500-600 DPS really too much for your DPS players?

If you look at the WoW Raid & Dungeons forums, you'll see a lot of newer guilds posting WoW Web Stats logs, and asking for help. And time and time again, you'll see the response focusing on the DPS players and how they are underperforming. People not packing enough +hit, using a bad spell rotation, using direct damage instead of DoTs.

If it was just a couple of people with problems, we could say that it was an issue with the player. But the scope of the problem is so large that I think it points to systemic faults in WoW.

DPS players have never had to reach 500 DPS before in the game. All solo mobs and quests are killable with much lower DPS. And it has to be this way, otherwise paladins or other low-DPS characters would never be able to kill anything. So for a DPS player, solo mobs die so fast that there is no need to tune her damage to the higher output.

Boss fights are also much longer than regular fights, and that means that Damage-Over-Time effects become more powerful than Direct Damage. A very common mistake is to see rogues using Eviscerate instead of Rupture. But this makes sense in regular play. Most mobs don't survive long enough for Rupture to finish. Instead, using Eviscerate to kill the mob faster is the way to go. The problem is that the rules change when it comes to boss mobs.

Another common mistake is not packing enough hit rating. There's a large jump in misses between level 72 and 73 mobs, especially for spells. Players who don't know about that jump often don't pack enough hit rating. They go for stats like crit, which have proved more useful in regular play.

The other major problem for DPS players is that they don't have enough feedback. They're just pouring damage into a central pool, and it is really hard to tell if your individual contribution is enough. In contrast, healers get immediate feedback. If a healer doesn't heal competently, people die. This forces healers to improve at a much faster rate. Similarly for tanks, if a mob gets away, or if DPS is consistently pulling off the tank, the tank knows she needs to improve.

So far, the common theme is that regular play, including 5-man instancing, does not prepare DPS players for raid boss fights. And when you combine that with the lack of feedback during boss fights, it's no wonder that DPS players are having problems, especially when first getting into raiding.

The only real way to improve DPS players currently is out-of-game research and theorycrafting. It would be better if there was a more organic, in-game way, to prepare DPS for raiding.

Monday, September 10, 2007

Time Management Article

Ciderhelm posted a superb article on the TankSpot forums: Time Management for Raiding Guilds.
For most people, thinking of the top tier of raiding guilds conjures thoughts of people who do nothing but play a computer game in a dark room somewhere. In fact, this is often not the case -- time for career, family, and friends are still very much a part of their lives.

The key is in time management.

It's very well written, and is very insightful. It also has tips on how a lower tier guild can take advantage of some of the time management techniques to improve their own progress.

Money Quote:
What players do not understand is how real a factor time is on the guild welfare as well as their own. Spending time gearing up for new content is almost never as well spent as time actually working on that content. This is especially true in the Burning Crusade, where stat differences and gear progression between Karazhan and Black Temple is relatively small; keep in mind that Nihilum cleared through Black Temple just 3 months after raiding began. Gear is good, but it is not key. [Emphasis mine]

Friday, September 07, 2007

Mages and AoE

If mages are supposed to be the kings of AoE damage, why do we always use warlocks in fights which require AoE?

Take Illhoof, for example. The standard strategy is to have a warlock spamming Seed of Corruption on Illhoof to take out the imps. But why not use a mage? Does it just come down to the fact that the warlock will be a lot easier to heal?

It seems a bit odd. If mages are supposed to be the best at AoE, they should be the first choice for fights which require AoE.

Thursday, September 06, 2007

More Healing Macros

I made and tested a couple of new macros tonight. I find that I am not very good about using short-cooldown trinkets and spells, and so I decided to try and come up with macros to use these automatically.

Holy Light Macro
#showtooltip Holy Light
/cast Divine Favor
/cast Divine Illumination
/script UIErrorsFrame:Clear()
/cast Holy Light

This macro attempts to cast Holy Light(Rank 11) with both Divine Favor and Divine Illumination. I'm a Flash of Light spammer, so when I need to pull out HL11, it's usually worthwhile to pop both these cooldowns.

Flash of Light Macro
#showtooltip Flash of Light
/use 13
/use 14
/script UIErrorsFrame:Clear()
/castsequence reset=10 Holy Light(Rank 5), Flash of Light, Flash of Light, Flash of Light, Flash of Light, Flash of Light

This macro attempts to pop both my trinkets whenever I cast my Flash of Light. This ensures that I get maximum usage out of my trinkets. The castsequence is so that HL5 keeps Light's Grace up fairly often.

You can adapt this macro so it works with any spell you spam (Fireball, Sinister Strike, etc.):
#showtooltip spell
/use 13
/use 14
/script UIErrorsFrame:Clear()
/cast spell

If you use these macros, you should go into Sound Options and uncheck Enable Error Speech. Otherwise it will drive you mad. Note that text error messages are cleared before you cast the actual spell. This is so you can see "Out of Range" or other useful errors.