We'd love to have Searing Totem use the new assist stance, but we felt it was too risky to just flip that switch without a great deal of testing. True pets have a control bar so it's possible to override their behavior if the default behavior is something you don't want. That's not the case for Searing Totem. We've just recently been able to get the totem to behave fairly predictably in a wide variety of situations – it has a lot of special case code designed to make it do what players want it to do. (Making an AI behave consistently isn't that hard; making an AI read players' minds is the hard part!) If assist works out well for pets, and some of the temporary guardians like Guardian of Ancient Kings, we'll turn it on for Searing Totem.
The Fire Elemental is even more complicated than Searing Totem because the totem is the master of the elemental, not the shaman. It generally works pretty well focusing on the Flame Shock target, but we plan on rebuilding the spell so that the totem summons the elemental (and killing the totem could still despawn the elemental), but the shaman is considered the master, which will solve some of the problems that arise.

Every Enhancement shaman loves seeing numbers fly across the screen when a huge multi-crit Windfury occurs, reinforced by our recent change to allow Windfury Weapon to trigger three additional attacks, rather than merely two. Windfury is ultimately a passive ability, though, and serves as an extension (albeit an awesome one) of your auto-attacks. In Burning Crusade, Windfury was prominent because shaman had very few active buttons to press, and long periods of downtime between them that some would fill by "twisting" totems. Not the most compelling gameplay. In Wrath of the Lich King, they arguably inherited the opposite problem, having so many buttons to press that there was never a free global cooldown, while no single ability felt particularly impactful or meaningful. For Cataclysm, we attempted to pare down the rotational complexity of the Enhancement shaman (removing the need to manually refresh Lightning Shield, removing Fire Nova from single target rotations, etc.) and at the same time created synergies that allow Lava Lash to do impressive damage.
As an aside, there were several questions that we didn't answer about whether Enhancement DPS is too low overall. It is, and we buffed it for 4.2: (http://us.battle.net/wow/en/blog/2723732)

As we've discussed in the past, any time a class has a meaningful portion of its damage only receiving 50% bonus critical damage, it's going to be difficult for crit rating to be an attractive stat. We'd certainly prefer if haste were a more attractive stat for Enhancement shaman. For many other melee classes, the great value of haste lies in its ability to increase resource generation. Enhancement shaman are not generally limited by any resource, so aside from more auto-attack damage and Windfury/Flametongue procs, haste currently yields more Maelstrom Weapon charges. We've thought about taking steps to make Maelstrom Weapon a more central mechanic for the Enhancement shaman spec, which would in turn make haste potentially much more valuable, but don't currently have any firm solution to announce.

Part of what makes the Enhancement shaman feel like a true hybrid is their even mix of melee attacks and magical ones, and changing the way spells like Lightning Bolt or Flame Shock work might have a serious impact on how Elemental shaman play. We do want Enhancement shaman using melee-oriented Agility weapons, though, and one solution we're considering is a mechanism that would make Enhancement shaman spells all scale from attack power, similar to what we did with Flametongue procs. In the meantime, we've taken steps (and will continue to take them) to ensure that while spell power weapons might at times, with certain gear setups, remain an interesting alternative, they aren't the strictly superior route.

For a long time we've resisted the temptation to add threat-reduction abilities to warriors and death knights because we don't want every class to have the mirror images of the same abilities. Class homogenization is a complex and philosophical discussion and probably worthy of a developer blog soon. In some cases, we realized that preserving flavor among classes was just holding classes back – reliable interrupts are a great example of this, where we finally just gave one to every melee and tank spec. But we feel like we have to remain vigilant about this sort of thing. While it might feel like a nice band aid if you are currently the character lacking what seems like a must-have ability, in the long term it can do harm to the game. You see a lot of players today who would rather trade some of their utility and possibly even game balance in order to have classes behave more differently from each other. You can disagree with that point, but it's hard to completely dismiss it and we certainly don't. Getting back to the original point, this may be one of those cases where we just need to give in and make sure all DPS specs have some sort of personal threat dump. We'd definitely want to do it as an active button that requires player interaction and not just a passive modifier that lets you just ignore threat as a game mechanic.

We'd dearly love to do this, and it's been something we've been working on, off and on for some time. The problem is that increasingly players place a really high and occasionally unhealthy emphasis on meters, and once there is an official Blizzard-supported meter, then that situation is only going to get worse. Anything that isn't portrayed in our meters with a great degree of accuracy is going to be misinterpreted and cause forum drama. For example, it's easy for DPS to inflate their meters on some fights by attacking targets that don't matter. How do we handle those situations -- trust players to know the difference? That's tricky, especially when the community has a penchant for distilling lots of fights down into a single measurement of DPS. As another example, the Restoration druid Tranquility is intended to fill a role similar to Power Word: Barrier or Spirit Link Totem. Yet the druid cooldown is an actual heal, which greatly inflates their meters to the extent that we see a lot of players complaining about how Resto druids are overpowered. Do we not show Tranquility on healing meters?
On the other hand, one benefit of having easy-to-use Blizzard meters would be getting players to focus on their own personal DPS instead of what the best players in the world are capable of. It makes developers cry when we see a good Fury warrior go Arms and do lackluster DPS just because they read that Arms DPS is higher. (Now, if that player just likes Arms or wants to try something different, more power to them.)
Also consider that damage and healing meters are valued by a pretty small set of the playing population as a whole. New UI features like the quest and equipment systems we added not so long ago, and even the upcoming Dungeon Journal, would be more widely used overall.
So the short answer is that it would be a very useful tool and we suspect we'll do it eventually, but we have an enormous responsibility to get it right, and even then it could do bad things to the community as a whole.

We generally introduce ramp-up time for two reasons. The first and most important is so players have a decision about when to switch targets. If there was zero cost for target swapping, then it would always be the right thing to do. We want to reward players, modestly, when they know when they should swap targets versus sticking with the original. The second issue is that ramp-up time helps us reduce burst in PvP.
The intent for Shadow Orbs was that procs weren't guaranteed so that there is some unpredictability involved to add gameplay. We could easily make it less random, but then they wouldn't be something you think about or factor in your rotation. Sometimes you won't get Shadow Orb procs and your DPS will be lower than it could be if you get really lucky. Shadow Priest DPS is balanced around the average of those two extremes. If you get lucky and get good procs, that's an unexpected bonus.
There's a thin line between something that's frustratingly random and something that is boring and has no gameplay. We have learned that when percent chances are too high, then rather than feeling like a bonus when it happens, it becomes very frustrating when it fails to happen.

The main reason we have the positional requirement is to have a different vector along which to design abilities. Backstab without a directional requirement could probably just be folded into Sinister Strike. It's a way to make abilities different from each other, in the same way we have ranged attacks vs. melee attacks, instant spells vs. cast time spells and physical damage vs. magic damage. Also consider that all melee should want to get behind a target, and it hurts all of their DPS when they cannot. We have also made the alternatives to the positional requirements much less of a DPS loss than they used to be. If you go into your Mangle rotation instead of your Shred rotation, your DPS will drop, but not catastrophically. Now there are some encounters where the positional penalty is just too extreme. In 4.2 we have the ability to make the "back" of a boss encompass 240 degrees, and we have done so for bosses like Magmaw, Sinestra and Ragnaros. Furthermore, there are fights where Killing Spree and Feral Charge just kill you. That obviously isn't acceptable. We have manually added some safeguards to try and manually solve a few encounters, such as Magmaw, but even that isn't bulletproof and we are investigating more robust and global solutions. But it's technically challenging given the diversity of our encounters.

As you probably suspect, we have a simple counter that measures the number of forum posts on a given class and we buff or nerf accordingly.
Seriously though, we look at a lot of different measurements, which becomes the full-time job of several designers. Our three most powerful tools are doing predictive modeling for how classes will perform under various scenarios and with various levels of gear, actually testing these numbers using characters in the game world, and then measuring the numbers generated by actual players on PTR or live servers.
Remember that we have access to a number of tools not available to players. While theorycrafters have gotten very good at reverse engineering how our damage calculations work, there are still a few opportunities where they get it wrong while we can just peek under the hood to remind ourselves how a calculation is made. Secondly, it's very easy for us to create a lot of characters with whatever gear we want and have them beat on whatever kinds of targets we want in a very controlled environment. We can also change any of the numbers to empirically test the outcome. Furthermore, we can automate character damage rotations to a much greater degree than macros can accomplish, which gives us an idea of the delta between theoretical maxima and more typical player performance (which includes things like human reaction speed, decision making and good old Internet lag).
The specific situation that the character is in matters enormously. Maximum sustained DPS is almost irrelevant in PvP when applying burst in controlled windows is king. Yet both numbers have a huge impact on the game and neither is more important than the other. In PvE, the specifics of an encounter can trump almost everything. We have very few Patchwerk-style fights these days, and sometimes we even buff or debuff characters directly as part of the encounter. Some specs are good on movement fights. Some do better when there is a lot of incoming damage. Some benefit from spreading dots. Some can shoot flying dragons. We tend to focus a lot of our balance effort on the current tier of raiding content, because that is what is most important to players, but even then we have to look at a wide variety of skill sets. We do look at scaling into future content, but we tend to obsess over it a lot less than players do, because we adjust classes quite often these days. We actually do read the forums a lot too, our own and all the others out there that you probably read, just to make sure there is nothing we're missing. Our community team helps enormously in this endeavor, particularly in helping to funnel the feedback from players from Latin America, Europe and Asia. We're in contact with expert gamers from around the world. We also all play the game a great deal and very often we personally catch a bug or something else that isn't working quite right.
As an aside, this is the kind of question we were really hoping to get more of with this series. It's open-ended, potentially interesting to a lot of different players, and not just a thinly veiled demand for buffs.

One of the defining elements of rogue gameplay is the feeling of building up potential against a single opponent, and then unleashing that power. Redirect allows rogues to "cheat" on these mechanics once per minute (or more often with Restless Blades), but if we removed these ramping elements entirely, rogues would lose a large piece of what makes them unique. We recognize that at the end of the day, however, many players would rather be powerful than unique. Ideally, we'd like you to be both. Having to build up combo points to operate at maximum effectiveness is a disadvantage compared to being able to just do maximum damage from the outset, and having to move to melee range to attack a new target is a disadvantage compared to being able to switch instantly from range. There's nothing inherently wrong with disadvantages, as long as they are counterbalanced by equally powerful strengths. Rogue damage recently has been lower than we'd prefer, even on fights with low movement and no target-switching, which should be absolutely ideal for rogues. We've taken steps in 4.2 to increase rogue damage output across the board, and we will continue to make adjustments until we feel that rogue performance is where we'd like it to be.

First off, DPS often do have important roles in fights, whether it's banging gongs for Atramedes or interrupting during the Nefarian encounter, or just knowing to run out of the dragon breath in the Drahga Shadowburner encounter in Grim Batol (since the healer can't possibly keep you alive through all of that). On any given encounter, we tend to give responsibility to a few DPS players instead of all of them, and we think that's ultimately a good thing. Not every player wants a ton of responsibility and we don't think it would be good for them, or the game, for us to force those players into high-pressure situations. It is a game after all – it's supposed to be fun. If challenging is what's fun for you, well, that's what Heroic modes are for. We think most players understand that taking on the healing or tanking roles is going to come with more responsibility, and those roles in turn tend to attract players comfortable or interested in having more responsibility.
Going even further, we'd say that one of the reasons our current raid encounters are considered so difficult is that the failure mechanics are fairly steep. We have a lot of "you're the bomb!" spells where if you fail to run out of the group, you can kill not only yourself, but the entire raid. That makes it harder to bring along inexperienced players or new recruits to see a boss encounter. Maybe those type of mechanics should stick to the Heroic modes of the fights, where everyone is presumably signing up for a lot of individual responsibility.

This is one of those interesting phenomena you observe when you do game design. According to our testing, Frost does comparable DPS today, and remains competitive in Heroic Firelands gear. Yet, Frost is much less popular than Fire or Arcane in the current raiding environment. Some of that could be tradition. Some of that could be that the DPS is close enough that players pick whichever mage rotation they enjoy the most. It's also possible that some aspect of our testing doesn't catch some factor that ends up suppressing Frost damage in the "real world" compared to our internal servers. That certainly happens sometimes.
It's hard to just look at logs and get an accurate picture of the mage DPS situation. When the best mages in the world are playing Fire and Arcane, it's natural to expect that there are a lot of huge DPS averages for Fire and Arcane mages. That might not mean that Frost damage is low, only that the best players aren't playing Frost. We see this sort of thing over and over again. As we mentioned recently, you can see Unholy DPS drop in 4.1 not because Unholy's DPS was nerfed, but because so many good DKs switched from Unholy to Frost. While it's ideal that all DPS specs are viable in all aspects of the game, and that remains our goal, it's a lofty goal. Frost remains the mage spec of choice in PvP, which is a better situation than if it were just a dead spec.
While some specs may do slightly higher damage than other specs within a given class, the differences aren't so great that you'd really be holding your group back if you played your favorite talent tree instead of the one with the highest DPS logs. In almost all cases, individual skill, gear, encounter specifics and Internet lag will have a bigger effect on your DPS than your spec choice (and often your class). Seriously, try Frost mages. Try Subtlety rogues. Their DPS is honestly pretty competitive.