Elitist Jerks Raiding Theory

 02/04/09, 2:54 PM #106 PSGarak Bald Bull     Garak Undead Warlock   Hyjal If you want to look at it as a statistical problem, it's actually very easy with a small number of assumptions. 1. Gather data on how geared people are when they first beat a particular encounter (or achievement or area). Not just an average, but a full distribution. In the simplest case, this is some weight of time spent practicing and average ilevel of gear, for a binormal distribution. 2. Find where on your distribution you tend to lie based on the encounters you have beated, probably in terms of standard deviations from the mean. 3. Find the corresponding point on the distributions for encounters you have not beaten. The average is easy to find, the spread can probably be found with some Bayesian analysis. The major issue at hand is that a dozen encounters isn't necessarily a solid enough data set to get good data, but maybe. Also, to make sure you're not gumming up your initial data, make sure that you only include data in (1) that has data available for (3), so you're comparing yourself to the same groups in both cases. This is one of those naive methods that ignores what you think actually matters and only looks at statistical performance. I have mixed feelings about that sort of theory, but they tend to work pretty well in a large number of domains. If you want, you can split fights apart by, say, DPS-checks, healer-checks, mobility fights, etc, but then you start having not enough encounters to find your average performance with any statistical significance.
02/04/09, 3:48 PM   #107
Douglas
Don Flamenco

Night Elf Druid

Earthen Ring
 Originally Posted by Faldrath But usually you're not facing cutting edge content. You're facing content for which there are raid guides, boss mods, EJ threads, etc. Then a raid leader will use that general experience I mentioned above to arrive at certain expectations when the raid faces something new. A good example is Sarth +3: discussion has established that you probably won't have a chance if your 25-man raid doesn't have enough DPS to kill Tenebron before the 2nd set of whelps.
Okay, this is getting towards the sort of thing I've been looking for.

For this case, okay, we can try such-and-such fight that we can win, and if we can win it in this particular way, it's a good indicator that we're ready to try this other fight. The linkage in this particular case is extremely direct though.

A list, knowledge-base, collection of those would be an awesome resource, no? Like, does "number of necrotic aura reapplications observed on Loatheb" correlate with success chance on any other fights? Or "overall damage taken by your DPS on Sapphiron"? Or "frequency of dance missteps on Heigan"? Or "total person-minutes your raid spends cursed on Lucifron" (for an old example)? Or even "the ratio of the time it takes to decurse your main tank to the time it takes your last DPS to get decursed"? What key indicators do we look for, and what do they mean?

Gah. Want more, and more transparent, metrics.

02/04/09, 3:52 PM   #108
solbergb
Don Flamenco

Gnome Mage

Earthen Ring
 Originally Posted by Douglas Maybe I'm incorrect about that experience being necessary. But how do you decide whether you're ready for Malygos if nobody in your group has ever seen Malygos?
Try it out. Wipe. Learn. Repeat.

Eventually you'll learn what to do with sparks. Pretty soon you'll want to stop them from getting to Malygos just to avoid his powerup. Then you'll learn they add to dps. You'll learn that you need a strategy to avoid most of the group from being breathed on after the vortex, and to keep health high due to vortex and other aoe damage.

Eventually you survive to phase 2. More trial and error as you work out what the discs do, and the bubble dynamics.

Once you make it to ph3, you'll know how far down the enrage timer you are, which gives you some idea how much dps you need to really have enough time in ph3. You'll learn about the stupid vehicle phase and somebody will remember that Aces High has a similar vehicle where people can practice.

Realistically though, only the first few guilds have to do it that way. Because of the internet, eveyrone else will have somebody on the raid who read a writeup somewhere so they have a basic idea what's going to happen. They'll still have to try it out to learn about tank survivability in ph1-2, raw raid dps in ph1 and all the ways to screw up the strat in all 3 phases that can lead to deaths.

But really..if you poke it and Malygos enrages in ph1 when the whole raid has survived, you don't have anywhere near enough dps. If you can't keep the raid or tank alive in ph1 when nobody is blowing up the raid you don' t have enough healing or sturdy enough tanks. The rest is just a learning experience for the raid on how not to screw up. This sort of thing you can pretty much get from wowhead. The damage Malygos does with his breaths is known, as is that of the adds in ph2, the area damage in ph1 from raw aoe damage+vortex etc. You should be able to gauge if that's too much for your healers based on past performance on other fights. His enrage timer and hitpoints are known, as are the limitations of the dragon mounts in PH3. That should give a sense of raid dps needed in ph1-2.

Ph1 is complicated by sparks. But you can look at something like Loatheb to get a sense of how good your crew is at taking advantage of those kinds of mechanics and assign some kind of multiplier to your patchwerk dps. PH2 seems to me the hardest to time, as it's highly dependent on your strategy for getting melees onto discs and how likely it is your ranged dps can actually target the stupid fliers while running around in bubbles. I can't think of another fight similar enough to that to do anything other than try it out and see how it goes.

02/04/09, 6:30 PM   #109
Montegomery
Aloof Aggravator

Sutiru

No WoW Account
 Originally Posted by Douglas Okay, this is getting towards the sort of thing I've been looking for. For this case, okay, we can try such-and-such fight that we can win, and if we can win it in this particular way, it's a good indicator that we're ready to try this other fight. The linkage in this particular case is extremely direct though. A list, knowledge-base, collection of those would be an awesome resource, no? Like, does "number of necrotic aura reapplications observed on Loatheb" correlate with success chance on any other fights? Or "overall damage taken by your DPS on Sapphiron"? Or "frequency of dance missteps on Heigan"? Or "total person-minutes your raid spends cursed on Lucifron" (for an old example)? Or even "the ratio of the time it takes to decurse your main tank to the time it takes your last DPS to get decursed"? What key indicators do we look for, and what do they mean? Gah. Want more, and more transparent, metrics.
The problem you're facing is that a "good indicator" is subjective and specific to your raid, your players. You could hypothetically compile a database of all guilds measuring these things, but ultimately you'd get the same generalizations you get anywhere else, with even less accuracy.

What you seem to want is, effectively, a purely binary answer to the question "Will we kill this boss if we attempt him now?" You're asking for a level of prognostication that no one but a veteran raid leader experienced specifically with your group of raiders can give you, from a tool which can't possibly understand the human elements in the system.

While the ideal of the resource you describe would be useful, realistically you're asking for the impossible. No matter how accurate we could make it you would still lack the concrete metrics you seek.

 Originally Posted by Vectivus ... you could very well have a concerto, but the closest the average listener gets to hearing it is the interpretation as put on by a group of small children with those little rainbow-coloured xylophones.
Monte's LoL Blog
Monte's LoL Stream

02/05/09, 12:14 AM   #110
jozga
Von Kaiser

Orc Warrior

Runetotem (EU)
 But how do you decide whether you're ready for Malygos if nobody in your group has ever seen Malygos?
Maybe I am missing something, but you are going to be ready for Malygos when you have cleared Naxx. Sure, it might be hard if you are moving fast but you are ready to try. There is generally a pretty linear progression route and things like the Sapphiron quest are usually in place to make sure you don't do it the wrong way. Further, the bottlenecks on the path are generally very apparent; bosses with a harsh enrage for example. If everything else is dead and Malygos is the next boss on the list then what else are you going to do?

 Bottom line...you can measure raid dps and you can observe deaths. Deaths will usually get better with practice, "glasschewing" a boss. If it isn't getting better with practice, look at your healing strategy or your overall strategy. Raid dps..you just look at a fight where the raid didn't die, compare it to the boss health and timers and say "damn, even if we did perfectly, we'd still lose". That usually requires more gear from the farm bosses, although something like waiting for heroism to come back online or swapping the buff mix can sometimes help.
This sums it up for me; 'glasschewing' is exactly right and this is how you figure out if you can beat a boss. The most transparent and useful metrics for estimating success are going to be the broader ones like 'number of deaths' and 'average dps'. Tools like WWS and Failbot help a lot here. I really doubt that this would highlight different players than the very specific metrics that Douglas suggests. Although I would suggest 'number of times died in a void zone at KT or Sarth' and 'number of times died on Thaddius' as pretty good moron checks.

02/05/09, 9:40 AM   #111
Rhaegal
Don Flamenco

Pandaren Monk

Zul'Jin
 Originally Posted by jozga Maybe I am missing something, but you are going to be ready for Malygos when you have cleared Naxx. Sure, it might be hard if you are moving fast but you are ready to try. There is generally a pretty linear progression route and things like the Sapphiron quest are usually in place to make sure you don't do it the wrong way. Further, the bottlenecks on the path are generally very apparent; bosses with a harsh enrage for example. If everything else is dead and Malygos is the next boss on the list then what else are you going to do?
So far in WotLK, Blizzard has done a good job with this, but that's not necesssarily historically accurate. Take the end of BC--yes, when Sunwell came out, guilds who had been farming BT for months were ready, but was a guild who had just downed Illidan for the first time? Unequivocally no. It remains to be seen how much Naxx/Malygos gear is "sufficient" for Ulduar, but a new fresh 80 guild at 3.1 who sweeps into Naxx and clears it probably won't be ready for Ulduar on the same reset.

My point is that there's always a "gearing up" stage between progression plateaus. Those plateaus may not necessarily be the end of one raid and the beginning of another--early bosses in one raid are often easier than the end bosses in the previous one. To use my previous example, the real "gear plateau" was Brutallus, not Kalecgos. The idea behind this is to find out whether you've spent enough time gearing up to reach the next plateau, instead of wasting a night or two of 10 or 25 people faceplanting on a boss they're physically not ready for, but didn't really know without trying.

Stand back! I'm going to try SCIENCE!

02/05/09, 11:14 AM   #112
sovelis41
Bald Bull

Draenei Shaman

Zul'Jin
 Originally Posted by Darian_TruBlade What you seem to want is, effectively, a purely binary answer to the question "Will we kill this boss if we attempt him now?" You're asking for a level of prognostication that no one but a veteran raid leader experienced specifically with your group of raiders can give you, from a tool which can't possibly understand the human elements in the system.
This is the most important thing to consider. I've played with a good majority of my raid group for the better part of 2+ years, so I'm very aware of what things certain players are good, or more importantly, not good at. Even if you're running a PuG, you should be able to get a good idea of what you can get away with a certain strat/boss after a couple attempts. This allows me to evaluate attempts on a new boss and pinpoint what is causing a wipe. When learning a new boss my priorities are as follows:

1) Keep the tank alive. A function of both healer capabilities and tank gear. This will allow you keep the fight going long enough to actually see any Important Mechanics(tm) even if you don't do the fight correctly.

2) Execute the strategy. This involves movement, positioning, assignment, and kill orders as needed.

3) Refine the strategy, Troubleshoot. If boss HP is still above 0%, go to Step 2, otherwise Go to Step 4.

4) Distribute epics, cheer wildly if boss was hard, laugh at Bob again for dieing, kill raid members with your opposite charge, and/or laugh at people looting badges in gas clouds/blizzards.

Step 2 and 3 are very important and will be the sole factor in answering the question: "Can we do this?" Answering it correctly takes some experience, and no amount of math modeling is going to answer it for you. Also, refining a strat doesn't end after a guild first. There is always room for improvement.

Alternatively, you could take the Homer approach: "Well, boy, you tried hard and you failed. The lesson here, never try."

[e] Not to take this too off topic, but a guild could most certainly progress in Sunwell after their first Illidan kill (yes, even Brutallus).

You pay for the whole chair, but you only need the edge.

02/05/09, 6:19 PM   #114
Allev
King Hippo

Tauren Druid

Mal'Ganis
 Originally Posted by Rhaegal Take the end of BC--yes, when Sunwell came out, guilds who had been farming BT for months were ready, but was a guild who had just downed Illidan for the first time? Unequivocally no.
I disagree. My guild had killed Illidan about a month before Sunwell came out, for the first time. We were definitely "ready" for the content at that level-- we needed to practice a lot on Kalecgos, but it was definitely an execution issue rather than a gear issue. Had the guild not collapsed, we were on track to hit the DPS benchmark of Brutallus around the same time that our healing execution was pulling together. We wouldn't have spent any more time improving our gear while on Brut than we had learning to click our portals. Or learning Illidan, or Archimonde, or Bloodboil, or Kael, or Vashj.

 02/22/09, 8:30 AM #115 Lavode Bald Bull   Lavode Night Elf Druid   Earthen Ring (EU) Delete
02/24/09, 7:39 PM   #116
Phlis
Don Flamenco

Draenei Shaman

Magtheridon
 Originally Posted by Douglas Gah. Want more, and more transparent, metrics.
Completely disagree here. Transparent metrics do not exist.

Just using the Sarth example, I completely disagree. As a raid leader I've never viewed 3D Sarth as a DPS race. It's entirely a survival fight. Once people are easily able to avoid flame walls and void zones, the fight is about adds. Having a tank, or two, extra to grab the adds keeps people alive, and, as opposed to other fights, 3D Sarth gets easier the longer you stay alive.

There is no key to knowing whether or not you can do any single fight. You need to try, and adjust. I think "on the fly adjustments" are a key part to successful raiding, whether during a fight or in between wipes, changing people's jobs, making minor switches(We killed M'uru after switching the sides our tanks were on, because one tank just needed that extra distance to get adds under control and the other didn't, and killed 3D Sarth after moving our Sarth Tank as far away as possible to give everyone else more room), deciding when(and who) to battle rez, all of the little active things are very important as a raid leader. Anyone can explain a fight or read a Bosskillers strat. Good raid leaders adapt to the situations put infront of them.

 03/06/09, 10:26 AM #117 Tya Glass Joe   Sylvain Night Elf Druid   Medivh hey I dont' really know where to ask that but since this thread was entitled "raiding theory" i thought that could be the place , feel free to indicate me if i'm wrong and plz where I could find my answer Here's my problem : I'm in a semi hadcore guild .We raid 12 hours a week, been clearing pretty much everything ( we're working on reg OS with 3 drakes alive atm). My question concerns group composition. We cover every buffs every raid, we got a good mix of healers, dps but our tanking team consists of 3 warriors and 1 feral druid. Recently , one of our warrior decided to play his hunter since we find that we lacked some diversity with our tankin team. Right now, we are facing two choices to promote one DK or to promote one Prot pally. The DK asked first a long time ago but some people are saying that anyway tank or dps there's no point of promoting him since he could bring the same buffs and will provide aproximatively the same amount of threat than the prot paladin while the prot pally will give our raid Greater Blessing of Sanctuary. Since I never played with either class as a MT or OT in a heroic raid, I'd like to have some insights on this very sensible question.
 03/06/09, 11:33 AM #118 Fqubed Fuck You Bed     Retired Blood Elf Paladin   No WoW Account (EU) If you are working on Sarth+3D you should take the DK. Simply he is a better tank due to the number of cooldowns he has available to him to stay alive for the double breaths. Raidwide Sanctuary is perhaps overrated. 3% less damage taken is not going to save a healer who pulled whelps and doesn't get them taunted off. If you plan on trying to get sarth+3d take whoever is the better player, equally skilled/geared the DK will most likely be a better tank for the encounter.
03/07/09, 4:45 AM   #119
Enova
Great Tiger

Night Elf Hunter

To be honest, you could get them both, if you think you could fit them in a rotation. In the context of Sarth+3, a DK will certainly be the ideal tank for soaking breaths while both Shadron and Vesperon are still alive, while a prot pally healing with Righteous Fury will make picking up the adds trivial. Apart from Sarth+3, both of them are otherwise perfectly fit for any other main or off tanking role in the game at the moment or in Uludar. Surprisingly enough, the only place where Blessing of Sanctuary will come in handy on Sartharion+3 is on the DK main tank. But the difference between a tanking and a dpsing DK is not in the buffs. Generally, a tank specced DK will do an okay job at dpsing given some decent offspec gear, but a dps specced one with decent tanking gear might fail miserably at tanking.

If I understand correctly, you're now running with three tanks (the feral and the two remaining warriors?). If so, rotating 5 tanks for (usually!) 3 spots a raid doesn't seem like an unreasonable idea.

 Originally Posted by XI- In summary, TBC raiding is easy. 9/10 encounters can be summarized with 1 phrase. Stay out of the fucking fire. If this is too difficult BWL was still there last I checked, so go have at it for some practice.
 Originally Posted by Kaubel You people are idiots
Guilty as charged ^

 03/07/09, 6:02 AM #120 Impeesa Glass Joe   Glutes Human Warrior   Laughing Skull Keep in mind, they already have a bear and it sounds like they're already working on 3 drake sarth, so I'm guessing they already have the MT situation for that fight covered. If that's the case, from a class standpoint it becomes a question of utility. I'm not really qualified to comment on that beyond the fact that either one would complement your existing tanks nicely as an AOE/offtank, but I will say that those considerations are probably overshadowed by the question of player skill and reliability. Ignore for a moment what class they play, and ask yourself which player you'd rather rely on.

 Elitist Jerks Raiding Theory