Related question: Now that all tanks have AoE tanking capability, what happens on trash and in add fights? Does any of it have to be controlled and single-target killed, or is it all just "gather it up and blow it away with AoE"?

If the latter, Mutilate is going to be a really painful spec to play :-(

It's easy to fix by having high dps mobs, meaning a tank would take too much damage if you don't control them.

Even if they don't... what's the problem with having lower personal DPS on trash as long as you're viable on bosses (where it really matters)?

Tanks have the threat needed to tank packs, but they won't always have the survivability to do it.

I still have 2 or 3 of Malacrass' adds CC'd, if only because it's not worth the DPS time and extra damage intake to burn them down.

Apparently you should still CC all but one of Faerlina's adds, with one getting slow DPSed and stopped within Execute range to be killed as soon as Faerlina enrages.

There are also times when you still want to focus fire on specific targets based on how dangerous they are. The Amani'shi melee mobs don't really pose a threat, but the Flamecasters still deserve to be burned down ahead of everyone else, as opposed to a Blizzard/Hurricane-fest that kills them all at the same time but results in much more AOE damage from the longer time-to-live.

Or, we could examine the distribution based on the number of possible roles per class:

1 role: Rogue, Mage, Warlock, Hunter.
2 roles: DK, Priest, Warrior.
3 roles: Shaman, Paladin, Druid.

So Vanquisher exchanges for sets belonging to seven roles, Conqueror for six, and Protector for six. Each token has one class with three roles and one with two. Each token has at least one each of melee DPS, ranged DPS, tanks and healers. DKs have been placed on a token that will have five sets without them - including a tank and a melee DPS - so the tokens remain balanced even if your guild hasn't yet levelled a DK to raid standard. Honestly, you don't get better balanced than that.

Well this is all true assuming that every class / role is divided equally. However, I doubt to see more than 1 moonkin, but 2 mages/locks/hunters are likely to be fairly common just to give an example.

The division should be based on the expected number of players of a class in a raid, not by role. Tokens aren't for specs, they are for classes. If it is expected that you'll have 1 bear, 1 moonkin and 2 resto druids, that should have a weight factor of 4/25, while a mage could have a weight factor of 2/25 to 3/25.

Only for that first kill. You also have to consider the longevity factor of a piece of loot, in which case classes with differing spec sets are going to have a much larger demand then classes with only 1 set.

For example, rogue/mage/hunter/warlock are going to have 1/3 of the longevity of druid/shaman/paladin, so the tokens of the former are left to rot on the boss while the raid still desperately wants more of the latter.

So tokens need to be balanced with sets in mind.

Well, remember that with dualspecc, off-specc-gearing could be something very needed for a guild. Say gearing up backup tanks/healers.

You're forgetting that each class can take a token for each of its specs. If it is expected that you have three Druids and three Mages in the raid, your first nine Druid tokens will have some value but only your first three Mage tokens won't be binned.

Let's say your raid composition is as follows:

Tank: 1 Warrior, 1 Paladin, 1 DK
Healer: 1 Paladin, 2 Priest, 2 Shaman, 2 Druid
Melee DPS: 1 Warrior, 1 Rogue, 1 Shaman, 1 DK, 1 Paladin
Ranged DPS: 3 Mage, 2 Warlock, 1 Druid, 1 Shaman, 2 Hunter, 1 Priest

You don't have a Feral Druid, but every other spec is present. Before your whole raid has a piece for each potential role and the tokens lose all value, you will need the pieces to drop this many times:

Vanquisher: 1x1 (Rogues) + 2x2 (DKs) + 3x1 (Mages) + 3x3 (Druids) = 17 tokens
Conqueror: 3x3 (Paladins) + 3x2 (Priests) + 2x1 (Warlocks) = 17 tokens
Protector: 2x2 (Warriors) + 2x1 (Hunters) + 4x3 (Shamans) = 18 tokens.

Very nearly perfectly even. The only token that is in any danger of going to waste too early is Protector, as that distribution has one more Shaman than may be expected. But even if you drop the fourth Shaman for another class, about the worst balanced distribution you'd be looking at is 20/17/15 (Enh Shaman dropped for Feral Druid).

Well, since its not exactly unlikely that you will indeed bring a feral over a 4th shaman, a range from 20-15 is pretty damn devastating.

It really isn't, you know. If we're getting two drops per kill, then given even distributions we're talking an average 22 weeks of kills before tokens begin to be wasted. Tiers of content are (hopefully) not going to last that long.

I don't care about the DPS numbers, what I care about is the exceedingly non-fun playstyle of tab-targeting, getting a single Mutilate off (on an unpoisoned mob so it hits like a limp noodle) and then having the mob die before I can use any finishers. Oh, and having my target knocked out of range by a mage or shaman AoE knockback is exceedingly awesomesauce too.

Hopefully Fan of Knives will help somewhat - do we know if it's still based on weapon damage so you have to carry swords around to macro in on AoE pulls? There was a blue post to say they'd normalised it, but reports from beta are inconsistent.

It's not like mobs are just going to all flop over dead after one hit. Most trash is easily AoE tanked and burned but almost all of it still has single target threats to prioritize.

How can you base drops in a world-wide game on what a small proportion of raid groups consider "expected"? Taking the example of our own little raid group, numbers have varied wildly over time - from 5 regular Mages back when we started Kara, to 1-2 regulars when we stopped raiding. We've rarely had more than 1 Warlock in a raid, frequently none. We've had 7 Shamans in one raid, 8 Druids in another (and that wasn't deliberate stacking, I assure you!) - and none or few of either of them in far more raids than that. And don't use the argument "yeah, but you're never going to get anywhere with a group like that" - we certainly weren't the most advanced guild, but that doesn't mean we're any more or less important that those who killed Kil'Jaeden, nor any more or less entitled to a given distribution of loot. If anything (and I no doubt risk censure here), groups such as this are more important as they're simply far more comonplace - and so make up a significantly greater proportion of the playerbase.

Division based on the number of different gearsets per token as as good a system as any, and, in my opinion, superior to one based on what would suit groups X, Y or Z

I certainly agree that they can't go on expected composition. While many guilds do min/max raid composition, from my experience in a more casual guild, we go with who we can get, which often has some weird compositions. Our SSC clear the other night had 0 mages and 1 warlock, but something like 5-7 each of paladins, shamans and warriors.

I do think the best way is to simply go with the two token solution, like in AQ40, but blizzard really seems reluctant to go back to that model.

What about a literally-role-based composition?

Have three tokens. One can be used for the tanking equipment of any class that can tank at all, and not for any other gear. One can be used for the healing equipment of any class that can heal at all, and not for any other gear. And one can be used for the DPSing equipment of every single class, and not for any other gear. The tank/healer/DPS proportion is something Blizzard assumes and tunes for, yes? So tune the drop rate accordingly.

(Maybe split the DPS one up into two tokens, if you're afraid of having their drop rate be so disproportionate.)

Yeah, a paladin might need one of each token in the long run if they want to collect all the sets. But is that a problem? Is that worse than "three of the same token"? Why is "tokens are for particular classes" better than "tokens are for particular roles"?

I think the RNG could be even more punishing on that setup.

The line between offspec and mainspec can become pretty blurry. Obviously if someone is taking a piece of loot that will never be used except in a grinding or PvP then from a raid utility standpoint those "offspecs" might be of little less worth.

However, I find its most often the case that classes that have multiple specs are often encouraged and sometimes even required to respec and change roles depending on raid composition. Such as "we're short a tank tonight, go respec Lasie" or "we've got enough tanks Lasie go respec DPS". For the greater benefit of the raid you need to have your hybrids pick up all the offspec sets they can, in order to provide greater raid flexibility if someone doesn't show up.

So in the case of a hybrid, off-specs are very important.

Because they can be clearly labelled as such.

You can place a "Class: rogue, paladin, warlock" mention on a token, and it's easily understood. I'm a rogue, I can roll/bid/whatever on this token. But "role: healer"? Watch the masses collectively scratching their head trying to decide "hmmm, does role tank mean there's a chance my class has a valid set with this token?" (it has not, but you'd be amazed at who can rationalize the fact that their mage is "tanking" in Maulgar's encounter as meaning they might have a "tank" set with lots of sta - that's what the raid lead ask them: "do you have your tank set?").

I can't think of any situation where this would be considered good advice. There are always some special mobs in pulls, that either cast spells, or do knockdowns/stuns/other cc on the tank or heal or just have more/less HP that you'd be better off killing over doing nothing. Even if it's a 5 random unexciting melee mobs-pull it's still faster for everybody if the single-target dps kills off one of two mobs before the rest fall since it lowers healer mana wasted and generaly speeds up stuff, as well as reducing the risk that some stupid aoe class will get aggro and die from a single hit, jumping around in melee like the idiot he is.

As an un-needed tank in instances (ZA, how I loathe thee and thy OT-requiring bosses) I still DPS even though it may be pitiful, and as a healer not needed to mainheal I keep everyone topped-off or just do some DPS. Anything beats just standing around.

I don't really understand what they don't like in the 2-token solution.
In fact, I don't understand why they don't simply make a SINGLE token solution. I never got the point of the multi-tokens idea, except creating potential waste.

Heck, they even let two items in set be buyable with badges, officially to "alleviate the RNG problem", so they are supposed to not want people wait too long to see their item set to drop. They could simply remove the possibility to buy them with badges, and make a single token system. Here you go, absolutely zero problem with RNG.

Maybe I missed something, but I don't see what's the point of purposedley making something painfully RNG, and then give a dubious makeshift compensation about the RNG on the other hand.

Its not too bad. Before if you had a dual pairing of either a Rogue, Mage, Warlock or Hunter in a grouping of three (i.e. Rogue, Mage, Druid) then tokens in that grouping would tend to be much less in demand in the long run due to the pure dps classes having no need for offspec gear.

Of course in the short term it would kind of suck if the drop rate was even, but as someone above mentioned if you up the drop rate of the larger grouping that should even things out.

Why have any loot determined by RNG? Why not just have "Illidan List A token" and "Illidan List B token" drop each kill instead of a random item from those lists? It would allow every one to get their DST/Skull/whatever that much faster, and have the drop be determined by player desire, not RNG. It actually would make far more sense to have each boss drop things that NPCs convert for you into a variety of different armor pieces or whatever desgiend specifically for each individual player than have a piece of armor drop that can be worn equally well by a Troll, Tauren, Human, or Gnome. It would lend more credance to the item actually being much more powerful than the random stuff you find in the world that you have to fiddle with to get to fit for your character.

In my mind, the real excitement is killing the boss; Blizzard wants part of it to be the anticipation of finding out what drops. To me, it only leads to depression when what I wanted didn't drop or worse something dropped no one could use. Maybe its just my personality though.

It would also mean that guilds' progression is only a function of how well their loot system gears them for progressing, not a function of RNG.